Long-‐term effects of small-‐scale spatial variations in fire severity on bilberry (Vaccinium myrtillus L.) growth in the boreal forests of eastern Finland
MEM Report (UNB) and M.Sc. Thesis (UEF) Laura Pekkola
Supervisors:
Dr. Jari Kouki (School of Forest Sciences, UEF)
Dr. Dave MacLean (Faculty of Forestry and Environmental Management, UNB)
January 26, 2016
Table of Contents
Abstract ... 2
Introduction ... 3
Fire History ... 5
Fire in Conservation ... 6
Biology and Importance of Bilberry ... 6
Research Problem ... 7
Materials and Methods ... 8
Study Site ... 8
Project Fire ... 8
Fire Severity Data ... 9
2014 Experimental Design ... 9
Vegetation Sampling ... 10
Analysis Methods ... 10
Statistical Analysis ... 12
Results ... 12
Discussion ... 16
Dwarf Shrub Biomass ... 16
Regression Analysis of Cover and Mass ... 16
Regression Model ... 17
Future Studies ... 18
Implications for Conservation ... 19
References ... 21
Abstract
Intensive forest management and the local fire history have contributed to the simplification of Finland’s forests. Negative consequences have been seen in plants, such as bilberry (Vaccinium myrtillus L.), which has been declining in Finland over the past 50 years. Bilberry is a socially, economically and ecologically important dwarf shrub in Finland, making it of interest in conservation planning. Prescribed burning has been suggested as a conservation tool in Finland to return species and structural diversity to the forests. Fire severity has been found to vary at small spatial scales within stands, but little research has gone into the effects of this variation on bilberry. The objective of this study is to determine whether i) dwarf shrub cover recovered at the stand level 13 years post-‐fire and ii) how small-‐scale spatial variations in fire severity affect bilberry growth 13 years post-‐fire.
Dwarf shrub and bilberry percent cover and biomass were sampled and compared with fire severity data from 2001 using a general linear model. It was found that dwarf shrub biomass at the stand level did return to pre-‐fire conditions by 13 years post-‐fire. It was also found that bilberry biomass and cover declined slightly with increasing fire severity. Therefore, fire at the stand level likely does not impact bilberry growth anymore after 13 years. However, more severely burned areas may still see an effect of fire at this temporal scale. Severe fires damage bilberry rhizomes, which decreases bilberry’s ability to grow by vegetative structures. Fire severity should be considered when planning to use fire for conservation, especially if it is an important area for bilberry.
Introduction
The structure, function and processes of forests are defined by the natural disturbances that occur within them (Lindenmeyer and Franklin, 2002). Though natural disturbances, such as fire, wind or insects, can seem to be damaging to forests, they are an important part of the natural cycle of the ecosystem. Disturbances create heterogeneity in age and spatial structure and allow for different species to establish, thus creating a variety of habitats and processes across the landscape (Bergeron et al., 2002; Lindenmeyer and Franklin, 2002). These effects caused by natural disturbances, as well as the patterns in occurrence are described by the natural disturbance regime of an ecosystem.
In the boreal forest, fire is an important component of the natural disturbance regime (Wein and MacLean, 1983; Esseen et al., 1999; Bergeron et al., 2002; Lindenmeyer and Franklin, 2002). Of the world’s 1.7 billion hectares (ha) of boreal forest, 5-‐15 million ha burns annually (Stocks 1991;
Kasischke and Stocks 2000; Conard et al. 2002). The area of forest that burns has, however, decreased in recent decades due to fire suppression practices intended to protect people, their property and timber resources (Niklasson and Granstrom, 2000). This has had negative implications on forests due to the importance of fires to the boreal ecosystem.
Fire is important in maintaining boreal ecosystem structure and function. Fire can maintain biodiversity by providing a variety of habitats across the landscape (Siitonen, 2001) and by allowing for deciduous tree regeneration, allowing for associated species to survive (Kouki et al., 2004). Fires also provide suitable growing environments for Scots pine (Pinus sylvestris L.), deciduous and other early successional tree species by decreasing the populations of later successional species, such as Norway spruce (Picea abies (L.) Karst), which tend to dominate landscapes that do not experience disturbances (Kuuluvainen, 2002; Similä and Junninen, 2012).
Fires maintain structural heterogeneity in forests by burning some areas severely, and leaving other areas untouched for hundreds of years. This helps preserve over-‐mature forests on the landscape, while allowing young forests to establish as well (Bergeron et al., 2002). This diversifies the available habitat on the landscape for organisms requiring different stages in succession. Fire can also expose mineral soil (Granström, 2001) and return nutrients to the soil (Mallik, 2003), allowing for favourable growing conditions for vegetation. These effects of fire, therefore, increase the overall diversity and function of the boreal forest.
Current forest management practices tend to simplify forest structure in terms of age, species and landscape distribution (Bergeron et al., 2002). Intensive, conventional forest management can also result in loss of fire-‐dependent understory species, loss of biodiversity in general and a simplification of the species present, with species such as spruce becoming overwhelmingly dominant (Lahti et al., 1991; Gustafsson et al., 1994; Östlund et al., 1997). A landscape of spruce alone reduces the variety in possible habitats that would exist if pioneer species and other mature tree species were present in higher numbers as well.
Though forest harvesting and fire both remove biomass from the forest, the end results are not identical (see e.g. McRae et al., 2001; Lecomte et al., 2006). Harvesting tends to remove all or most of the trunks, whereas fire leaves behind much of the large wood, instead removing fine fuels such as branches (Sippola et al., 1998). Fire also provides an element of disturbance that harvesting and site preparation cannot in terms of exposing mineral soil and returning nutrients to the forest floor. Therefore, there is a crucial difference between harvesting practices and fire.
Finland has a high proportion of managed forests, with 90% of forested land under management for timber production (Kouki et al., 2001; Löfman and Kouki, 2001). Therefore, forest management decisions that are made in Finland affect a large proportion of the country’s forests.
It has been suggested that the main threats to Finland’s forests include habitat fragmentation and scarcity of natural fire (Raunio et al., 2008; Rassi et al., 2010). In order to address these threats, forest management in Finland could incorporate natural disturbance emulation. The purpose of this form of forest management is to imitate natural disturbances in order to optimally maintain the processes and structures created by such events while still allowing for resource extraction.
Wind, snow, water, insects, mammals and fire are the important natural disturbances found in Finnish forests (Similä and Junninen, 2012). Fire in particular can be emulated in forest management in boreal forests (Kuuluvainen 2002; Lindenmeyer and Franklin, 2002). New forestry models have been introduced to minimize the negative effects of intensive management in forests (Kohm and Franklin, 1997; Larsson and Danell, 2001) by retaining trees within cutblocks, thereby creating cutblocks that are more naturally shaped and sized while maintaining landscape-‐level variability through appropriate planning of cutblock placement. Prescribed burning can also be incorporated by pairing it with harvesting. This reduces the growth of fast-‐
growing species such as grasses, allowing for establishment of forest species (Johnson et al., 2014), and maintains structural diversity in the forest (Lilja et al., 2005).
Fire also creates diversity by burning at different severities within the stand (Dyrness and Norum, 1983; Miyanishi and Johnson, 2002). This was mainly affected by the moisture of the lower moss and humus layers of the forest substrate. Most research on the effects of fire severity have been done on tree recruitment (e.g. Johnstone and Chapin 2006; Johnson et al., 2014). Fire can have varying effects on tree seedling establishment and germination, however this is not the interest of this study.
To date, there has been little research on the effects of this small-‐scale variation on understory vegetation. One study looked at the patchy nature of fire severity in Sweden and found that less severely burned areas within the stand acted as refuges for bryophytes (Hylander and Johnson, 2010). Therefore, there is a need to look at the effects of small-‐scale variations in fire intensity on other understory vegetation as well.
The purpose of this paper is to examine the effects that variation in fire has on understory vegetation, in particular bilberry (Vaccinium myrtillus L.) in the boreal forest of eastern Finland.
A better understanding of the effects of fire on bilberry will help in evaluating the efficacy of forest restoration practices in Finland as measured by a non-‐timber forest product.
Fire History
Understanding the fire history of a region can provide information on the natural occurrence of fire as well as the historical effects of anthropogenic activities. In the boreal forests of Sweden, fire history better correlates with human cultural history than it does climate data (Niklasson and Granström, 2000; Lehtonen and Huttunen, 1997), meaning humans have played a large role in controlling fire occurrence on the landscape. Due to the geographical and cultural similarities within the Nordic countries, it can be expected to be similar in Finland.
Naturally, the fire return interval in eastern Finland’s forest has been estimated to be in the range of 80-‐200 years, depending on the forest site type (Pitkänen and Huttunen, 1999; Haapanen and Siitonen, 1978). Prior to the mid-‐17th century, fires tended to be few in number and large in size throughout the Fennoscandian boreal forest (Niklasson and Granström, 2000). This suggests that large natural fires that were greater than 1000 ha historically burned through large tracts of forest with little influence from humans.
In the 17th century, increased human influence in Fennoscandian boreal forests became apparent in the change in fire pattern. Fires became more numerous and smaller, and fire return intervals shortened significantly (Niklasson and Granström, 2000). At this time, slash-‐and-‐burn agriculture became prevalent on the landscape, as humans cleared forests for fields. From the 1600’s through to the mid-‐1800’s, fires became very common in Fennoscandia due to slash-‐and-‐burn agriculture (Heikinheimo, 1915). With this form of agriculture, the fire return interval was only 40-‐50 years (Lehtonen, 1998).
The prevalence of slash-‐and-‐burn agriculture began to decrease due to the establishment of the Finnish Forest and Parks Service in 1859 (Lehtonen and Huttunen, 1997). The Forest and Parks Service began the practice of fire suppression in Finnish forests to protect the timber resource, and to allow for the growth of forestry as an industry. Since the implementation of fire suppression, the fire return interval has been 80-‐120 years (Granström, 1996; Parviainen, 1996;
Haapanen and Siitonen, 1978). However, size of fires remains smaller than what was seen under the natural disturbance regime.
Between 2000 and 2012, an average of 566 ha of forest burned in Finland annually (Finnish Forest Research Institute, 2013). Forest fires that do occur in Finland tend to be quite small, measuring approximately 1 ha in size (FAO, 2006; Finnish Forest Research Institute, 2004). Therefore, fire suppression has limited the frequency of fires to more natural levels, but has not reinstated the large fires that were seen in the 17th century.
Fire in Conservation
Reintroduction of fire has been considered as a conservation tool in Finnish forests. In Finland, 4.8 million ha of forest (13% of the land area) is protected or has some level of harvesting restrictions present (Finnish Forest Research Institute, 2013). In these areas, prescribed burning is being tested as a conservation tool for forest restoration (Vanha-‐Majamaa et al., 2007; Similä and Junninen, 2012). This could return some of the benefits and processes that fire provides to the landscape in a controlled manner. In managed forests, most of the biomass is harvested for the timber industry. Therefore, even if fires do occur, some of the benefits associated with mature trees and fire do not occur in the forest.
One of the major benefits of fire is that it is an important creator of heterogeneity due to variations in fire characteristics. There are three important characteristics of fire (Bergeron et al., 2002): frequency, how often a fire occurs in an area; intensity, the energy output per unit length of fire front (Byram, 1959); and severity, the transfer of heat into soil and degree of organic material removal by fire (Rowe, 1983). The factor of interest in this study is severity. Factors that can influence the severity of a fire include topography, weather and fuel load and type (Johnson et al., 2014). Therefore, as a fire passes through an ecosystem, it will leave behind areas that are severely burned, as well as unburned areas that are refuges for surviving species.
When severe fires pass through forests, it is the understory vegetation that is most affected. As fire severity increases, more heat is transferred into the soil, resulting in more damage to roots and rhizomatous structures. For example, bilberry rhizomes – underground vegetative reproductive structures – will die within 10 minutes of exposure to temperatures above 50°C (Schimmel and Granström, 1996). Fire severity can, therefore, affect how species such as bilberry regenerate after fire.
It has been found that rhizomatous species (e.g. Vaccinium dwarf shrubs) regeneration post-‐fire is negatively correlated with the depth of burn (Schimmel and Granström, 1996). Since rhizomatous growth is a slow process, Schimmel and Granström (1996) hypothesized that it would take many years for bilberry to regenerate in a severely burned area.
Biology and Importance of Bilberry
Bilberry is a dominant forest floor species in Scandinavian boreal forests that produces berries that are picked for consumption and that feed wildlife (Atlegrim and Sjöberg, 1996). It is a rhizomatous dwarf shrub that is very common in mature stands of both pine and spruce in eastern Finland. Bilberry is an important component of boreal forest function due to its abundance, contribution to ecosystem services, and economic and cultural importance in Finland (Kettunen et al., 2012). Therefore, conservation practices in forests should consider not only the trees, but also other ecosystem components, such as dwarf shrubs.
Bilberry plays a role in providing food for species such as capercaillie, bank voles, moose (Selås, 2001; Lakka and Kouki, 2009; Selås et al., 2011); red deer (Hegland et al., 2010); roe deer (Mysterud et al., 1997); and pollinators (Rodriguez and Kouki, 2015). As a member of the dwarf shrub functional group, bilberry also plays a role in soil carbon and nutrient cycling and driving boreal ecosystem dynamics through its role as an important food source (Nilsson and Wardle, 2005; Kolari et al., 2006).
It is estimated that Finland’s forests produce approximately 182 million kilograms of bilberries annually (Turtiainen et al., 2007), of which 5-‐6% are gathered for personal use or sale (Turtiainen et al., 2011). In 2012, 6.8 million kg of bilberry were brought to market, producing a revenue of 12.2 million euros (Finnish Forest Research Institute, 2013). There is also a high cultural involvement in wild berry picking, with approximately 60% of the population of Finland picking berries annually (Saastamoinen et al., 2000).
In Finnish National Forest Inventories over the past 50 years, dwarf shrub cover, including that of bilberry, has declined from around 50% to below 30% (Reinikainen, 2001). This is partially due to the damage caused to the rhizomes by clear-‐cutting and mechanical site preparation (Atlegrim and Sjöberg, 1996; Tolvanen, 1994; Hautala et al., 2001). Clearcuts are also damaging to the bilberry plants due to the drying caused by the increased exposure to heat from the sun in open areas (Mäkipää, 1999). Fertilization practices can also negatively impact bilberry, even years after application (Strengbom and Nordin, 2008).
Therefore, bilberry occurrence in Finnish forests is declining, due in part to conventional forest operations. Bilberry is a significant component of boreal forests, so considering it in forest conservation and restoration practices is important in order to conserve the related ecosystem services. Due to the cultural, socio-‐economical and biological importance of bilberry, its perennial growth form as well as its abundance in the boreal forest of eastern Finland, it was chosen as the study species.
Research Problem
In this study, the effects of small-‐scale spatial variations in fire severity on bilberry were studied.
As a rhizomatous species, the regeneration post-‐fire depends partly on the degree of damage to the rhizomes during fire (Turner et al., 1998; Macdonald, 2007; Schimmel and Granström, 1996;
Pidgen and Mallik, 2013). Landscape-‐scale studies of fire effects on understory vegetation have been conducted previously (e.g. Johnson et al., 2014; Marozas et al., 2013; Schimmel and Granström, 1996), but the effects of small, within-‐stand variations have not been researched for bilberry. This study will examine effects 13 years post-‐fire, which is a longer temporal scale than most fire follow-‐up studies.
Fire severity varies within stands, with severely burned areas being dispersed among more lightly burned areas (Kafka et al., 2001). This has been measured through the variations in humus consumption by fire (Chrosciewicz, 1976; Dyrness and Norum, 1983; Zasada et al., 1983). Change in humus depth has been found to be an acceptable indicator of heat transfer belowground (ie.
fire severity) (Schimmel and Granström, 1996). Fire severity was the fire characteristic chosen for comparison in this study, as this information was readily accessible and quantifiable.
The first objective of this study was to examine the average recovery of dwarf shrubs on a burned site, 13 years post-‐fire. It was expected that dwarf shrub biomass would recover to levels close to pre-‐fire levels in 2014. The second objective was to determine whether small-‐scale variation in fire severity affect bilberry growth. It was expected that areas of the site that were more severely burned in 2001 would exhibit lower bilberry growth in 2014 due to damages to rhizomes.
Materials and Methods
Study Site
The study was conducted in Patvinsuo National Park, located in Lieksa, Finland (at 63°N, 30°E), which falls at the transition between the south and middle boreal vegetation zones (Ahti et al., 1968). The average annual temperature for the region is 2.4°C, with the hottest month, July, averaging 16.5°C and the coldest month, January, averaging -‐10.5°C. Average precipitation is 620mm for the year, of which 279mm falls between May and August (Pirinen et al., 2012).
The forests in the region are dominated mainly by Scots pine and Norway spruce. The forests in the region have a history of intensive management, though the forests within Patvinsuo National Park have been protected from harvesting since the park opened in 1982. Prior to the 20th century, most of the forests in Finland, including the eastern region within which this study took place, were subject to slash and burn agriculture (Pitkänen et al., 2002).
Understory vegetation within the sites consists mainly of mosses, lichens and dwarf shrubs, with lingonberry (Vaccinium vitis-‐idaea L.) and bilberry being the dominant dwarf shrubs. The high cover of dwarf shrubs, mosses and lichens indicates that the forests fall under the classification of cowberry-‐crowberry, or Empetrum-‐Vaccinium type (EVT) and bilberry, or Myrtillus type (MT) (Cajander, 1949). The dominance of pine in the forest of our study sites indicates a drier forest type, which supports the classification under EVT.
Project Fire
In 1999, 24 sites were selected in the area of Ilomantsi and Lieksa to form Project Fire (Kouki, 2013). Project Fire is headed by Jari Kouki of the School of Forest Sciences at the University of Eastern Finland. The sites were selected in mature (150 year-‐old) pine-‐dominated stands, with a mixture of spruce and birch. Each site is 3-‐4 ha in size, which corresponds with the average clear cut size in Finland at the time. Three replicates of 8 treatment types were created: clearcut, 10
m3 retention (ie. 10 m3 of living trees were left on site after harvest), 50 m3 retention, and unharvested, with each harvesting type being either burned or unburned. In 2000, an initial structural inventory was conducted, encompassing information on: living and dead trees, other vegetation composition, organisms, herbivory, and soil.
Between November 2000 and January 2001, the 18 sites requiring experimental harvesting were harvested. The following summer, on June 27-‐28 2001, the 12 sites requiring fire disturbance were burned. In the years following the burning, monitoring of short-‐ and long-‐term effects of the treatments has continued.
Fire Severity Data
To obtain a measure of fire severity, humus layer depth measurements were taken before and after the burning done in 2001 by Jarkko Laamanen for his master’s thesis (Laamanen, 2002).
Humus depth and mass were measured approximately every 10 m over an area of one hectare both before and after the fire. Interpolated surfaces for both before and after burning were created using the GPS positions of the humus depth measurements. Comparison of these allowed for fire severity variation to be estimated for each study site. Higher fire severity was denoted by a higher percent decrease in humus layer thickness.
In general, the site had quite low fire severity (Laamanen, 2002). The humus depth on the site chosen for the 2014 study decreased on average from 46mm to 42mm. The fire severity was likely low due to the lack of fuel, since there was no slash from harvesting on the unharvested site. The days the prescribed burning occurred weather conditions were clear with winds at approximately 2.3m/s. Relative humidity was between 41-‐47%. These conditions were deemed good for the prescribed burning, though a stronger wind would have increased fire severity (Laamanen, 2002).
2014 Experimental Design
One unharvested-‐burned site (Site 30) was chosen from the 24 study sites. Only one site was chosen because this study was intended as an exploratory descriptive study to determine trends in bilberry growth after fire. Data were collected in June and July 2014, which falls within the main growing season of the vegetation of interest.
Within each site 100 plots were laid out in a grid with 10 m spacing, as close as possible to plots that were measured in 2001 after the fire. A handheld Trimble GPS unit with the 2001 plot data was used in combination with a tape measure to lay out the plots. The coordinates were recorded
at each point using the Trimble unit. These coordinates were used to compare the locations of the bilberry samples with the fire severity measurements from 2000.
GPS coordinates for each plot were recorded for a period of 2 minutes, with a measurement interval of 5 seconds. Measurements were taken in this way to increase the accuracy and precision of the GPS position, as small errors in location are significant when comparing to the fire severity information from the earlier study. This method allowed for error of the measurements to be taken into account, allowing for an accurate position to be obtained. GPS data were transferred and viewed as a shapefile in ESRI’s ArcMap 10.2.
Vegetation Sampling
At each of the 100 plots, a 30 cm diameter ring was used to delineate the vegetation sample area.
Data on percent vegetative cover and biomass of bilberry was collected at each site. Percent cover was estimated visually to the nearest percentage of the area of the sample circle that was covered by bilberry, and recorded. The percent coverage for other functional groups (e.g.
mosses, grasses, herbs, dwarf shrubs, seedlings, lichens, etc.) were also visually estimated to the closest percentage and recorded.
All aboveground biomass originating within each circle was cut and separated into bags by functional group. In good weather conditions, plants were separated into functional group in the field. This was ideal, as separation after air drying (as was done when collection occurred in poor weather) was time-‐consuming and potentially damaging to the samples. Bilberry biomass was separated from the dwarf shrub biomass.
All biomass was taken back to the university for drying and weighing. Biomass was dried at the University of Eastern Finland in ovens at 60°C for 72h. The samples were then moved to a desiccator to cool prior to weighing. Oven dry weights were recorded for all samples.
Analysis Methods
2001 humus depth data from before and after the fire for the study site were added into ArcMap 10.2 as two separate point files. 2014 bilberry mass and cover data were added as two separate point files as well. The humus depth data was interpolated to create a surface of predicted humus depths using the Interpolate tool.
Interpolation allows for data values to be predicted for an area, using a collection of sample points. The spatial relationship between data points can be used to estimate a surface of values
from a group of points. In our case, humus depth measurements were taken 10 m apart, but we wanted to know estimated humus depth at points between the actual samples.
Kriging was chosen as the most appropriate method of interpolation for the humus depth data.
Kriging assumes that distance between points reflects spatial autocorrelation that can be used to explain variation in the values of the created surface. In our case, it follows that the fire would have burned with more similar severity in points closer together than in points farther away from each other, making kriging the best option.
Table 1 shows the parameters that were input for the kriging of both the before and after fire humus depth data files. These were taken directly from Laamanen (2002), where a spatial autocorrelation for humus depth was determined.
Table 1 -‐ Parameters used in kriging function performed in ArcMap 10.2 to predict humus depth surface for Site 30 both before and after fire in 2001. Parameters were taken from Laamanen (Table 13, 2002)
Type Lag Size Major Range Partial Sill Nugget
Before Spherical 10 39 180 95
After Spherical 10 42 300 75
The bilberry mass and cover data was overlaid on the before and after fire humus depth surfaces (see Figure 2). The Extract Multi Values to Points tool was used to extract the humus depth before and after fire from the two interpolated surfaces at the bilberry data points from 2014. This resulted in two new values being added to the attribute table of the bilberry data file (which included both mass and cover data).
Percent change in humus depth was calculated from the estimated values of before and after fire humus depth that were extracted. This gave an estimate of the percent change in humus depth due to the fire in 2002 at each bilberry data point measured in 2014.
This method of acquiring humus depth percent change due to fire has high error associated with it. Surfaces created by kriging have error associated with them, because the values across the surface are estimated from surrounding existing data points. Error increases as distance between data points increases. Because our percent change in humus depth was calculated from two estimated surfaces, the error in these values is possibly double that of the humus depth surfaces themselves. This explains why some of the points at which percent change in humus was calculated ended up with a negative value.
The negative values were removed from the dataset prior to statistical analysis since it does not make sense for humus depth to increase during a fire. A possible alternative might have been to change negative values to 0 for analysis, but due to the large number of points with negative percent difference, it was thought that this would probably bias results. Removal of the points ensured that areas of very low fire severity were not over-‐represented in the data, which could have been the case if negative values were simply changed to 0. This, however, likely resulted in low fire severity to be under-‐represented in this study.
Statistical Analysis
A two-‐tailed t-‐test was used to evaluate the difference in average dwarf shrub biomass (g/ha) before the fire, after the fire and in 2014. This gave an average indication of recovery of the site in the 13 years since the fire.
Multiple linear regression was used to analyze the data set gathered in 2014. Separate models were run for the response variables (Y) bilberry cover (%) and bilberry mass (g), with the general form of:
𝑌 = 𝛽%+ 𝛽'𝑋')+ 𝛽*𝑋*)+ 𝑒)
where X1 was percent decrease in humus depth and X2 was month of bilberry sampling. The interaction between percent decrease in humus depth and date was found to be insignificant in both regressions, so the model was simplified to the one shown above.
X2 was a dummy variable, where X2=0 when the sample was taken in June and X2=1 when the sample was taken in July. This allowed for the effect of sampling date to be taken into account when looking at the effect of percent change in humus depth on bilberry mass/cover. Samples taken in July were in general heavier and had higher cover because the growing season was further along. This is to be expected, and not of much interest in our study.
Results
Dwarf shrub biomass was negatively affected by fire, with biomass decreasing from 862.4 kg/ha to 251.1 kg/ha due to the fire in 2001 (Figure 1). By 2014, dwarf shrub biomass was 1093.1 kg/ha, which was significantly higher than the 2001 after fire measurements (p<0.01; Figure 1). In fact, by 2014 dwarf shrub biomass had slightly surpassed the biomass from before the fire. This relationship was found to be marginally significant (p=0.06).
Figure 1 Dwarf shrub aboveground biomass in 2014 (n=83), before fire in 2001 (n=33) and after fire in 2001 (n=29). The letters
above the bars indicate statistical significance, with ‘a’ being statistically significant from ‘b’ at α=0.05. Data for Before and After Fire are from the thesis written by Laamanen (2002).
Figure 2 shows the percent cover and mass of bilberry overlaid on the fire severity surface. The larger circles and darker colours of the circles on the map indicate where bilberry percent cover (red) and mass (green) were highest. These spots tend to correlate with the lighter grey areas of the fire severity surface, where percent decrease in humus depth was lowest. The smaller circles tend to land on areas with higher fire severity (darker grey/black). Therefore, with visual interpretation of this map, some relationship between fire severity and bilberry growth can be seen.
The regression analysis resulted in the following equations:
𝑀𝑎𝑠𝑠 = 0.53 − 0.0065𝑋') + 0.30𝑋*) (1) 𝐶𝑜𝑣𝑒𝑟 = 0.97 − 0.010𝑋')+ 0.40𝑋*) (2)
The regression indicated that sampling month and percent change in humus depth explained 14%
of the variation in bilberry mass (R2 = 0.14, F2,57 = 7.66, p = 0.0011; Figure 3). Both percent decrease in humus depth and date were both statistically significant factors in the regression (p
≤ 0.05). A 1% decrease in humus depth predicted a 0.65% decrease in bilberry mass (equation 1). For both months, as percent change in humus depth increases, the mass of the bilberry plants sampled decreases.
The sampling month and percent change in humus depth explained 17% of the variation in bilberry percent cover (R2 = 0.17, F2,57 = 5.819, p = 0.005; Figure 4). Both percent decrease in humus depth and date were both statistically significant factors in the regression (p ≤ 0.05). A 1
% increase in humus depth resulted in a 1% decrease in the percent cover of bilberry (equation 2). These results indicated that areas of higher fire severity in 2001 had lower bilberry plant mass and percent cover in 2014.
Figure 2 Bilberry percent cover (red) and mass (green) measured in 2014 overlaid on the humus percent decrease (fire severity) map
Figure 1 Relationship between log10 bilberry mass and percent decrease in humus depth with date of sample collection in 2014
as a factor: June: Date=0; July: Date=1. Bilberry mass is slightly higher in July than June. Higher fire severity, indicated by higher percent decrease in humus depth, resulted in lower bilberry mass (R2 = 0.18, F2,57 = 7.66, p = 0.0011).
Figure 2 Relationship between log10 bilberry cover and percent decrease in humus depth with date of sample collection in 2014
0.00 0.25 0.50 0.75 1.00
0 20 40 60
Percent decrease in humus depth
Log 10 bilberry mass (g)
Date 0 1
0.0 0.5 1.0 1.5
0 20 40 60
Percent decrease in humus depth
Log 10 bilberry cover
Date 0 1
Discussion
Dwarf Shrub Biomass
Dwarf shrub biomass in 2014 was similar to the aboveground biomass that was present in 2001 before the fire, at about 850-‐1000 kg/ha (Figure 1). 2001 pre-‐fire and 2014 biomass were both significantly higher than the biomass after the fire, which was about 250 kg/ha. Therefore, aboveground dwarf shrub biomass was significantly reduced during the fire (Laamanen, 2002).
It also indicates that the biomass on the site as a whole had enough time to recover in the 13 years between the 2014 measurements and the fire.
The biomass in 2014 was slightly higher, though not significantly, than what it was in 2001. This may be due to dwarf shrubs growing better due to the fire, reduced competition from other species, increased nutrient load due to fire or differences in measurement protocols between the study in 2001 by Laamanen and ours in 2014.
In a study with a shorter time frame, lingonberry (Vaccinium vitis-‐idaea L.) had not yet returned to pre-‐fire conditions by 4 years post-‐fire (Marozas et al., 2013). Looking at a longer time period, it was found that lingonberry had lower percent cover on burned sites than it did on control sites, even 14 years post-‐fire (Parro et al., 2009). Finally, a study done on the Project Fire sites found that in 2011, 10 years post-‐fire, lingonberry had not returned to pre-‐fire percent cover (Johnson et al., 2014). These results suggest that the 13 years of our study were not long enough for the dwarf shrubs to return to pre-‐fire conditions.
Comparing between these different studies is difficult, as the fire severity and method of burning can be very different. In the study by Parro et al., the measurements for unburned control were taken 14 yeas post-‐fire, just in areas of the stand that had not been burned. Environmental interactions in the burned stand may have effects on even the unburned areas, potentially making these plots not ideal for comparison to pre-‐fire vegetation conditions.
Also, the other studies that were referenced measured lingonberry only, whereas our dwarf shrub component included also bilberry, heather (Calluna vulgaris L.) and potentially other species. Though all of these species fall into the dwarf shrub functional group, it is possible that they react uniquely to fire, and cannot be assumed to behave as lingonberry does in fire.
Regression Analysis of Cover and Mass
The regression analysis shows a significant decreasing trend in both bilberry cover and mass with higher percent decrease in humus depth (ie. more severe fire). This implies that areas that were more severely burned in 2001 had slightly lower bilberry cover and mass in 2014. Therefore, it could be suggested that as fire severity increases, bilberry growth is negatively affected, even 13 years post-‐fire.
The explanatory power of the model, however, is not very high, as can be seen by the low R2 for both mass and cover. This is typical in biological systems where many factors can affect a single response. The model used in this study is quite simple, and does not capture all of the sources of variation that may influence bilberry growth post-‐fire. Other factors that should be studied include: patterns with co-‐occurring species, soil moisture/nutrient conditions and variability between sites. Therefore, the conclusions drawn from these results cannot be relied upon with high certainty. More research will be required to obtain definitive results pertaining to fire severity and bilberry growth.
Previous studies have generally found a slower recovery time for bilberry as fire severity increases. For example, in a study looking at bilberry recovery 5 years post-‐fire, bilberry had not recovered at all (Schimmel and Granström, 1996). Bilberry is a sprouter, meaning it revegetates using the rhizomes in the ground. It has been found that sprouters tend to have lower cover after fire as fire severity increases, due to the effects on underground rhizomatous structures (Wang and Kemball, 2005).
In a study on slash burning, it was found that bilberry recovery post-‐fire was very slow, with cover not increasing substantially in the first 10 years after fire (Ruokolainen and Salo, 2006). This supports our conclusion that 13 years may not be long enough for severely burned areas to recover.
Though the results of our study were not definitively conclusive due to the low explanatory power of our model, previous studies do support the trends we found. Other studies have found that high fire severity can be harmful to bilberry, and it can take many years for bilberry to recover. Therefore, there is support to our hypothesis that higher fire severity within a fire may affect bilberry more negatively.
The vegetative structure of bilberry, however, may influence how it is affected by small-‐scale variations in fire severity. If vegetative structures of bilberry connect aboveground plants over a large area, then damage to rhizomes in a small, specific location may have negative impacts on the biomass and cover of that entire organism. Bilberry rhizomes can extend up to 5m underground (Maubon et al., 1995), meaning underground structures can connect aboveground structures over quite a large area. Therefore, some of the noise in the model may be due to the non-‐localized effects of rhizome damage to aboveground bilberry growth. Studies looking at the impact of rhizome damage to the growth of aboveground structures of bilberry should be conducted.
Regression Model
There is a slight difference in the intercepts of the regressions between 2014 biomass measurements that were taken in June compared to July. The cover and mass of bilberry was less in June than in July, which is logical with the progression of the growing season, explaining the different intercepts. This is not really of interest for describing the effects of fire severity on bilberry growth, except that it should be taken into account.
Including data on the other species that were growing with bilberry could be helpful in strengthening this model as well. The forest floor of our site was completely covered by vegetation, meaning that there was no empty space for any of the species growing there.
Therefore, looking at interactions between bilberry and the other species and functional groups of plants that grew in the area may give indications of why bilberry cover and mass were what they were. It is possible that post-‐fire, bilberry was outcompeted by other species, preventing it from reaching pre-‐fire conditions.
Including multiple sites would account for inherent variability in environment that is linked to sites. This would likely improve the fit of the model, as other environmental factors that could influence bilberry growth post-‐fire would be taken into account.
Future Studies
Future studies should therefore expand on the number of sites that are measured, the vegetation that is analyzed and take into consideration other environmental variables (microtopography, soil conditions, etc.). The scope of this study was not large enough to consider all of these factors, but they can influence the growth of bilberry and should be accounted for.
The fire severity data used in this study had some error associated with it as well. The points of measurement were too far apart to capture small-‐scale variations accurately (Laamanen, 2002).
The errors in calculations of humus percent decrease during the fire also point to the inaccuracy of the fire severity data (Figure 2). In the future, measurements for fire severity would ideally be taken closer together, and before and after measurements would be taken from the same location. This would reduce the error associated with comparing two interpolated surfaces.
Also, having pre-‐fire bilberry biomass and cover measures would give a more definitive conclusion of the effects fire has on bilberry growth after fire. With before and after measurements taken regularly over the time period after the fire, more accurate bilberry response information could be gathered. This can help in understanding the full long-‐term response of bilberry to fire, which is important to ecological processes and services that may rely on bilberry.
For expanding the scope of this study from effects of fire on bilberry to applications in forest management, many other factors need to be considered. Harvest treatment, fertilization, thinning practices and site preparation for planting are all part of the process of preparing a stand for use as a timber resource. These activities all have effects on the understory vegetation of the forest stand, and must be considered when looking at the effects of forest management on species such as bilberry.
Combining prescribed burning and harvesting, as would occur in managed stands, was studied on the study sites in Lieksa. It was found that the combination of the two disturbances had a more negative effect on understory vegetation than either disturbance alone (Johnson et al.,
2014). Therefore, the combination of harvesting and prescribed fire may be harmful to understory plants such as bilberry.
Fire, however, can help reduce the thick humus layer that exists in northern boreal forests. This humus layer can hinder the regeneration of trees, making fire a useful tool to use after harvesting (Parro et al., 2009). Therefore, the importance of considering multiple perspectives when planning forest management goals is apparent, and it can be difficult to reach a solution that is beneficial for all.
Current forest management includes harvesting methods such as clearcuts, which have a homogenizing action on the forest. Clearcuts decrease the heterogeneity of the forest stand by simplifying the age structure of the developing stand, and by removing dead wood components from the stand (Kuuluvainen et al., 1996). Different methods of forest management create different successional paths in the forest (e.g. Uotila and Kouki, 2005), meaning the decision of what forest management to implement is important for determining the community that will exist on a site in the future.
Fire is only one tool available in forest conservation and restoration practices. The above-‐
mentioned practices can also be modified through forest management to have lesser impact on the surrounding environment while still fulfilling resource acquisition goals. Future studies should look at the combined long-‐term effects of fire with harvesting, site preparation, fertilization and thinning practices on bilberry.
Implications for Conservation
The descriptive results of this study in Northern Karelia, Finland in combination with previous studies, suggest that fire severity in pine-‐dominated stands of the boreal forest has a negative relationship with bilberry cover and biomass 13 years post-‐fire. Sites with higher fire severity had lower cover and mass of bilberry. This could have some implications on the recommended use of fire in forest restoration and conservation practices in Finland.
Bilberry is a significant component of Finland’s forest biodiversity, ecology and processes (Turtiainen et al., 2011; Kettunen et al., 2012). Therefore, understanding how fire affects it is important in designing restoration and conservation practices. There are many other values in the forest that need to be considered as well, such as timber, wildlife, recreation, ecosystem services and human health. Therefore, goals for forest restoration and conservation have many perspectives to consider, with bilberry being only one of them. Below are recommendations for the implementation of forest conservation practices with bilberry as a key component of the goals.
The most important consideration for bilberry in forest conservation is the fact that it has been on a decline in Finland over the past decades. Therefore, conservation practices should focus on reversing this decline. Fire is an important tool in the management of forests for multiple uses, but its utility for the conservation of bilberry may be limited.
