Post-fire species composition and regeneration of understory vegetation in a boreal forest in central Sweden

Anna Hassel

Student

Degree Thesis in Ecology 30 ECTS

Master’s Level

Report passed: 12 January 2018

Supervisor: Johan Olofsson

Abstract Post-fire survival, composition and regeneration of understory species in the boreal forest have shown to be affected by several factors, where consumption of the organic soil layer together with altered soil properties play important parts. There has however also been shown that the pre-fire site characteristics affect the post-fire understory vegetation. This study aimed to investigate the effects of fire and pre-fire site characteristics on understory regeneration and composition at a local scale in a boreal forest. Classification of species richness of the understory species together with measurements of biomass in terms of leaf area index (LAI) and normalized difference vegetation index (NDVI) were performed in a Pinus sylvestris forest in the Gärsjön catchment area, three years after a stand-replacing wildfire. Data of site index, fire severity on soil and moss, fire severity on shrubs, stand age, and remaining humus depth were also used. A total of 36 species of vascular plants (10 forbs, 14 graminoids, 5 dwarf shrubs, 2 ferns, 1 shrub and 4 trees) together with 3 species of bryophytes were recorded in the area. The study revealed that understory species composition was explained by remaining humus depth and site index. The regeneration of the understory was affected differently, where LAI was affected by site index, and NDVI was connected to both site index and fire severity on soil and moss. LAI and NDVI differed in their sensitivity in capturing differences among plant species, where higher values of LAI were associated to species such as E. sylvaticum, P. erecta, C. arundinacea and J. conglomeratus, while NDVI was related to both the ground and field layer, with high values associated to a high abundance of C. canescens and C. ovalis. According to my result, it can be concluded that NDVI is a more appropriate measure of post-fire re-establishment and recovery of understory vegetation in the boreal forest. Keywords: boreal forest, fire, understory, species composition, LAI, NDVI

Table of Content

1 Introduction ............................................................................................................................. 1

1.1 Background ........................................................................................................................... 1

1. 2 Aim and research questions ................................................................................................. 2

2 Material and methods ............................................................................................................. 2

2. 1 Study area ............................................................................................................................ 2

2.2 Measurements and study design .......................................................................................... 2

2.2.1 Understory composition ................................................................................................ 3

2.2.2 Understory regeneration ............................................................................................... 3

2.3 Statistical analysis ................................................................................................................ 4

2.3.1 Species composition ...................................................................................................... 4

2.3.2 Understory regeneration ............................................................................................... 4

3 Result ....................................................................................................................................... 4

3.1 Species composition ............................................................................................................. 4

3.2 Understory regeneration ...................................................................................................... 8

4 Discussion ............................................................................................................................... 9

4.1 Species composition ........................................................................................................... 10

4.2 Understory regeneration .................................................................................................... 10

5 Conclusions ........................................................................................................................... 12

1

1 Introduction 1.1 Background Wildfire is one of the most important drivers of vegetation dynamics in boreal forests (Kasischke, Christensen and Stocks 1995; Angelstam 1998; Kasischke and Johnstone 2005). The understory vegetation (tree seedlings, shrubs, forbs, mosses and lichens) is affected by fire in several ways, depending on the severity of the fire (Neary et al. 1999; Certini 2005; Gongalsky and Persson 2013). Fire severity is a variable used to describe and define ecological impacts in a burned area, and is often measured by visual inspection of charring and consumption of aboveground plant structures, or by consumption of the organic soil layer (i.e. depth of burn, e.g. Schimmel and Granström 1996; Ryan 2002; Shenoy, Kielland, Johnstone 2013).

Aboveground structures of the understory vegetation are often consumed regardless of fire severity, predominantly due to their exposure when it comes to height and absence of protective structures (Schimmel and Granström 1996). Effects on the belowground structures of the ground vegetation (roots, buds, rhizomes and seeds) vary depending on the consumption of the organic soil layer (Van Wagner 1983; Wikars and Schimmel 1999). Due to a cold climate and slow decomposition rates, the organic soil layer in boreal forests is typically deep and compact (Wang and Kemball 2005), and the degree of consumption is mainly dependent on moisture content and burn duration (Kasischke and Johnstone 2005). Schimmel and Granström (1996) have shown that understory regeneration depend on heat transfer into the ground, where maximum temperatures reached are mainly a function of organic soil consumption by the fire. Understory species that regenerate from buds and rhizomes will therefore benefit from a light or moderate fire severity, where regenerative structures will be able to sustain maximum temperatures (Bond and Keeley 2005; Boiffin, Aubin and Munson 2015). A high fire severity may start germination of seeds by exposing them to surface conditions when the overlaying organic layer is consumed, or kill seeds and reduce the seed bank. Colonizing species from surrounding areas will be favored by a substantial consumption (high fire severity) of the organic soil layer, which exposes the mineral soil and open ups the canopy (Bond and Keeley 2005; Boiffin, Aubin and Munson 2015). Survival, regeneration, and post-fire composition of understory species are therefore highly dependent on depth distribution of regenerative structures in the soil (Parro et al. 2009). The understory is further on affected by fire-induced alterations of soil properties. Severe fires lead to a reduction of moisture-holding capacities in soils, since consumption of the organic soil layer reduces both soil temperature through decreased insulation, and soil moisture content through increased evaporation and reduction in plant water uptake (Neary et al 1999; Certini 2005; Smithwick et al. 2005). A high heat transfer into the soil may also decrease or remove invertebrates and microbial communities important for decomposition processes, which may change nutrient cycle rates and thus reduce the nutrient availability (Neary et al. 1999; Certini 2005). A reduction in moisture-holding capacities in soils may also slow down recolonization of invertebrates and microbial communities, which further on affects the nutrient availability for understory vegetation (Neary et al. 1999). Changes in nutrient availability occur as an effect of volatilization, increased erosion and leaching (Certini 2005; Shenoy, Kielland and Johnstone 2013). These changes in soil properties may affect both productivity of the burned area and species composition of the understory vegetation (Nilsson and Wardle 2005; Mack et al. 2008). The impacts of the above-mentioned effects do, however, vary. It has been shown that a fast recolonization of fire-disturbed areas can prevent nutrient losses and changes in soil properties (Certini 2005), while others state that the impacts may vary considerably due to the diverse changes in abiotic and biotic factors following a fire (Neary et al. 1999; Smithwick et al. 2005).

2

Despite the importance of fire severity for understory regeneration and composition in boreal forests, several studies have indicated that there may be other variables affecting the understory to a higher degree than fire severity (Boiffin, Aubin and Munson 2015; Day, Carrière and Baltzer 2017). For example, pre-fire characteristics such as site index, basal area, climatic conditions and stand age have shown to be important in determining in post-fire understory composition. Further, the primary species that can colonize a fire area have been shown to play an important part in directing changes in species composition in the longer run (Day, Carrière and Baltzer 2017).

1. 2 Aim and research questions The main aim of this study is to investigate the effects of fire and pre-fire site characteristics on fire-induced changes in understory regeneration and composition at a local scale in a boreal forest. The questions that will be addressed are: 1. Is understory regeneration affected by fire and pre-fire site characteristics? 2. Is understory species composition connected to understory regeneration, fire and pre-fire

site characteristics? To answer these questions, vegetation regeneration was studied by non-destructive in situ measurements of leaf area index (LAI) and normalized difference vegetation index (NDVI), together with inventories of understory species composition and species richness, three years post-fire in Hälleskogsbrännan nature reserve in central Sweden. Effects of additional factors, such as fire severity, post-fire humus depth, stand age, and site index on post-fire species composition and understory regeneration were also considered.

2 Material and methods

2. 1 Study area The study was conducted in the Gärsjön catchment area, a 2250 ha area in the northeastern parts of Hälleskogsbrännans nature reserve, situated north of Ramnäs in central Sweden (59°53’ N, 16°8’ E). During July-August year 2014, a stand replacing wildfire took place in the area where around 13000 ha of forest land burned (Västmanland County Administrative Board, 2015 a). Before the fire a major part of the Gärsjön catchment area was production forest, and therefore consisted of a mixture of clearcuts, monospecific and even-aged forest stands (although the age between stands differed) (Västmanland County Administrative Board, 2015 b). This study has been concentrated to the upland areas of the catchment dominated by Pinus sylvestris L. Shallow and nutrient poor soils are dominating the upper parts, with more nutrient rich and thicker layers in the gradient zones. Pre-fire vegetation was dominated by Calluna vulgaris, Empetrum nigrum, Vaccinium vitis-idea L., Vaccinium myrtillus L., with mosses such as Hylocomium splendens, Pleurozium schreberi and in the drier areas lichens such as Cladonia spp. (MSB 2015).

2.2 Measurements and study design Prior to this study, measurements of fire severity and remaining humus depth were conducted across the catchment area of Gärdsjön, one year post-fire. These measurements were performed in circle plots with 10-meter radius, placed along a grid with 300-meter distance in between. There were 41 sampling points distributed within each circle, 10 at each meter and compass bearing (N, E, S, W), with one additional sampling point in the center (Fig. 1). Fire severity was inventoried by visual inspection of the consumption and charring of plant and moss structures together with the consumption of the organic soil layer, and was classified into four severity levels. The fire severity on the organic soil and mosses and fire severity on shrubs were classified separately as 0 = no impact, 1 = lightly scorched, 2 = no

3

living moss/shrubs, 3 = totally consumed. Additional data of pre-fire stand age and site index have been provided by previous land owners, where site index is a measurement describing site productivity (Hedwall et al. 2013). The additional measurements of LAI, NDVI and inventory of understory composition conducted for this study were performed in the same plots as previous measurements, with a total of 55 plots for LAI and 27 plots for NDVI. Due to practical inconveniences, NDVI measurements could only be performed in half of the plots.

Figure 1. Plot design showing the 41 sampling plots within on circle as black dots, distributed at each meter and compass bearing (N, E, S, W). Understory composition was measured at every second plot, resulting in 21

sampling plots per circle.

2.2.1 Understory composition Understory composition was measured using the point-intercept method (Jonasson 1988; Hudson and Henry 2009), using a 50-cm point tripod frame with five pins (ø<1 cm) regularly distributed. This method assures an objectively inventory of the species composition (Godall 1952; Jonasson 1988). The pins were passed vertically through the understory vegetation layer and each time a plant touched a pin that species was recorded. The ground layer was only registered once per pin. Inventories of species were repeated at every second sampling point within the circles, resulting in 21 measurements and 105 pins per circle. Polytrichum commune and Polytrichum juniperinum occurred in the fire area, but were often entangled and grew closely together, which made it hard to identify the specific species; they have therefore been aggregated as Polytrichum spp. Measurements were sometimes hindered by large amounts of branches and leaning or fallen tree trunks, and some sampling points were therefore excluded.

2.2.2 Understory regeneration Understory regeneration was measured with LAI and NDVI, three years post-fire. Both measurements were used in order to catch potential variation in vegetation height. LAI and NDVI have shown to be strongly correlated (Wang et al. 2004; Tagesson, Eklundh and Lindroth 2009; Song 2012) and are in this study used as a proxy for understory regeneration in terms of biomass. NDVI have shown to get saturated at higher values of LAI (>3, Metzger et al. 2016), decreasing the correlation between the two variables. Measurements of LAI at Gärsjöns catchment area did, however, never exceed an LAI value of 2 (appendix S1), which therefore excludes the problem of saturation. LAI (defined as the area of leaves per unit of ground area) is closely linked to ecosystem processes such as respiration and transpiration (where higher values of LAI is coupled with an increase in ecosystem processes) and is used as an indicator of primary productivity (Law and Waring 1994; Gower et al. 2001; Lindroth et al. 2008). LAI of the understory was

4

measured by incoming PAR (photosynthetic active radiation) above and below understory canopy using an AccuPAR LP-80 device, which allowed non-destructive measurements of the vegetation. LAI was measured in all 41 sampling points within the circles. Similar to inventories of understory composition, measurements could sometimes not be performed due to a large number of branches and stems. NDVI, defined as the difference between absorbed PAR and reflected near infrared light of the vegetation, gives a measure of ‘greenness’ or vegetation density (Stenberg et al. 2004; Lukes et al. 2016). This makes NDVI a more fitting measurement when it comes to lower vegetation heights (Payero, Neale and Wright 2004), where non-destructive LAI measurements can be hard to obtain. NDVI was measured using a SpectronSense2 data logger (Skye Instruments Ltd.) mounted on a hand-held pole with an incident sensor and light reflected sensor attached. A cosine correcting head was used in order to diffuse incoming light. NDVI was measured in all 41 sampling points within the circles.

2.3 Statistical analysis All statistical analyses were performed using R 3.4.1 (RStudio Team, 2015).

2.3.1 Species composition The species composition data was summarized at plot level and visualized using nonmetric multidimensional scaling ordinations (NMDS; Kruskal 1964), making it possible to see differences and similarities between the occurrence of species. The NMDS analyses were conducted using the metaMDS function in the vegan package (Oksanen et al. 2016). To further interpret the species composition, fire severity on soil and mosses, fire severity on shrubs, post-fire humus depth, stand age, site index, LAI and NDVI were added as predictors to the ordination space with the function ‘envfit’, in order to interpret how they are correlated to species composition three years post-fire. Only significant relationships between environmental variables and species composition were plotted. Since NDVI was measured in fewer plots than was LAI, an individual NMDS was performed in order to see the correlation between NDVI and species composition.

2.3.2 Understory regeneration The NMDS revealed that fire severity on soil and moss was strongly correlated with fire severity on shrubs and post-fire humus depth in both LAI and NDVI plots (Pearson’s r=0.68, -0.67, 0.71 and -0.77 respectively). Fire severity on soil and moss was thus positively correlated to fire severity on shrubs in both LAI and NDVI plots, showing that shrubs were consumed with a high consumption of the organic soil layer and present mosses. Fire severity on soil and moss was further on negatively correlated to the remaining humus depth, where an increased consumption of the organic soil layer resulted in a shallower post-fire humus depth. These correlations were expected, since these variables are highly interconnected. Hence, fire severity on soil and moss was chosen as a predictor variable in multiple linear regression analyses due to a higher amount of data. Potential effects of fire severity on soil and mosses, stand age, and site index on LAI and NDVI plots were analyzed with multiple linear regression models, where mean values of each variable on plot level were used. LAI and NDVI were used as response variables and the remaining measurements were used as predictors. LAI was log transformed in order to ensure homoscedasticity and fit the assumptions for parametric analyses.

3 Result

3.1 Species composition The Gärsjön catchment area was affected by a high-severity stand-replacing wildfire. Inventories one year post-fire revealed that the fire severity on soil and moss was high in the

5

plots where LAI and NDVI were measured, with a median of 2.73 (IQR=0.30) and 2.73 (IQR=0.37), respectively (Table S1, Fig.2). This means that the consumption of the organic soil and mosses was high, and was totally consumed in a majority of the plots. As expected, a high fire severity on soil and moss was also directly related to a high fire severity on shrubs and remaining humus depth (Fig. 2, detailed site characteristics in Table S1). Pre-fire stand age varied a lot among plots due to the burned area being used for production forestry prior to the fire, where the median stand age was 57 (IQR=74) and 49 (IQR=49) for LAI and NDVI plots, respectively. Site index, describing pre-fire site productivity for P. sylvestris, had a median of 22 (IQR=4.5) and 22 (IQR=6) for LAI and NDVI plots, respectively. The median of LAI was 0.37 (IQR=0.28) and the median of NDVI was 0.69 (IQR=0.12, Table S1).

Figure 2. Distribution of fire severity classes for soil and moss together with fire severity classes for shrubs in plots where LAI has been measured (A and B) and where NDVI has been measured (C and D). Observe that y-axes values for A and B differ from C and D.

Although some species have successfully established in the Gärsjön catchment area three years post-fire, the density of the field layer (73 hits/100 pins) and the cover of the ground layer (54 hits/100 pins) are still low. A large part of the ground was covered by litter (24 hits/100 pins) and stones (12 hits/100 pins), and a smaller part consisted of bare ground (9 hits/100 pins). In total 36 species of vascular plants (10 forbs, 14 graminoids, 5 dwarf shrubs, 2 ferns, 1 shrub and 4 trees) together with 3 species of bryophytes were recorded in the area (Table 1). The most abundant vascular plant species were perennial forbs such as Epilobium angustifolium (21 hits/100 pins) and Trientalis europaea (2 hits/100 pins), together with graminoids such as Deschampsia flexuosa (12 hits/100 pins), Carex pilulifera (3 hits/100

B A

C D

6

pins) and Carex canescens (2 hits/100 pins). Also, commonly occurring were the horsetail Equisetum sylvaticum (4 hits/ 100 pins) and dwarf shrubs, mainly Vaccinium myrtillus (4 hits/100 pins) and Calluna vulgaris (4 hits/100 pins), together with early-successional trees such as Betula spp. (8 hits/100 pins) and Populus tremula (3 hits/100 pins, Table 1). In the ground layer, Polytrichum spp. (38 hits/100 pins) and Ceratodon purpureus (15 hits/100 pins) were the most abundant moss taxa. The liverwort Marchantia polymorpha (<1 hit/100 pins) was less abundant (Table 1). Table 1. Summary of species abundance in the field- and ground layer in three-years post fire, displaying number of hits per 100 pins. Species are listed from most abundant to least abundant.

Field layer Growth forma Mean number of hits per 100 pins

Epilobium angustifolium F 20.54

Deschampsia flexuosa G 11.47

Betula spp. T 8.27

Calluna vulgaris DS 3.97

Vaccinium myrtillus DS 3.90

Equisetum sylvaticum F 3.83

Populus tremula T 3.19

Carex pilulifera G 2.71

Carex canescens G 2.37

Trientalis europaea F 2.16

Salix spp. T 1.94

Calamagrostis arundinacea G 1.41

Vaccinium vitis-idaea DS 0.87

Rumex acetosella F 0.86

Luzula pilosa G 0.84

Molinia caerulea G 0.69

Carex echinata G 0.69

Potentilla erecta F 0.66

Pinus sylvestris T 0.55

Calamagrostis purpurea G 0.53

Pteridium aquilinum F 0.53

Rubus idaeus S 0.32

Senecio sylvaticus F 0.16

Eriophorum vaginatum L. G 0.14

Schoenus ferrugineus G 0.14

Juncus conglomeratus L. G 0.12

Carex pallescens G 0.07

Vaccinium uliginosum DS 0.07

Arctostaphylos uva-ursi DS 0.07

Lathyrus pratensis L F 0.07

Carex ovalis G 0.05

Lotus corniculatus F 0.05

Festuca ovina G 0.04

Galium uliginosum F 0.04

7

Rhinanthus minor F 0.02

Ranunculus acris F 0.02

Ground layer Growth form Number of hits per 100 pins

Polytrichum spp. B 38.10

Litter

23.98

Ceratodon purpureus B 15.03

Stone

12.09

Bare ground

9.35

Marchantia polymorpha B 0.89

Root

0.55 a= Growth from: B, bryophytes; DS, dwarf shrubs; G, graminoids; F, forbs; T, trees; F, ferns; S, shrub.

The NMDS analysis of the plots where LAI have been measured revealed that species composition was significantly correlated (p<0.05) with site index and post fire humus depth and that it could explain LAI (stress=0.19, k=2, Fig. 2). The first axis (NMDS1) can to some extent be explained by LAI and post-fire humus depth, where increases in the variables align with low NMDS1 values (Fig. 2). This shows that species such as C. echinata, P. erecta, C. arundinacea, J. conglomeratus and E. sylvaticum are correlated to higher values of LAI and humus depth. Humus depth is further on negatively correlated with E. vaginatum L. and “root”. Site index increases with elevated values of the second axis (NMDS2) and is positively correlated with species such as C. ovalis and P. aquilinum, and negatively correlated to the graminoid C. canescens (Fig. 2).

Figure 2. Ordination plot displaying the result of the NMDS analysis including the plots where LAI was measured, showing the correlation of understory species composition and environmental variables, three years pots-fire. Environmental variables are represented by arrows.

8

The NMDS analysis further on revealed that NDVI was significantly correlated with species composition (stress=0.17, k=2, Fig. 3). The NDVI gradient is somewhat connected to increasing values of the NMDS2 axis. Species positively correlated to the gradient are the grasses C. ovalis and C. canescens, while it is negatively correlated to “root” (Fig. 3).

Figure 3: Ordination plot displaying the result of the NMDS analysis including the plots where NDVI was measured, showing the correlation of understory species composition and NDVI, three years pots-fire.

3.2 Understory regeneration The multiple linear regression analysis for LAI revealed that a higher site index, which is an indication of pre-fire site productivity, could explain 15% of the variation in LAI (Table 2, Fig. 4), indicating that a higher pre-fire site productivity is connected to higher LAI values. There were no significant relationships between the remaining variables, meaning that the variation in LAI is independent of variation in fire severity and stand age (Table 2). Table 2: Result of multiple linear regression models displaying the effects of fire severity and pre-fire site characteristics on LAI and NDVI.

Note: Level of significance is **: 0.001 and *: 0.01.

9

Figure 4: Result from the multiple linear regression, displaying the positive relationship between LAI and site index.

The multiple linear regression analysis for NDVI showed that 37% of the variation in NDVI could be explained by site index and fire severity on soil and mosses, and by the interaction of these two variables (Table 2, Fig 5). The interaction term revealed that a fire severity over 2.6 has a stronger negative effect on NDVI in sites where pre-fire site index was lower (Table 2, Fig. 5). An opposite effect can be seen at a fire severity lower than 2.6; however, this result is based on only a few data points and therefore be treated with caution.

Figure 5. Result from the multiple linear regression displaying the relationship between NDVI and fire severity on soil and moss at different values of site index (SI).

10

4 Discussion

4.1 Species composition High severity fires lead to a substantial consumption of the organic soil layer, exposing the mineral soil, and opens the canopy (Bond and Keeley 2005; Boiffin, Aubin and Munson 2015), which creates beneficial regeneration conditions for a diversity of species. The vegetation inventory in this study revealed that 14 graminoids and 10 forbs occurred in the fire area three years post-fire, making them the most abundant growth form. The dominance of graminoids and forbs aligns with previous studies, which have indicated that they become more abundant after high severity fires (Kayes, Anderson and Puettmann 2010; Shenoy, Kielland and Johnstone 2013), and where graminoids have seen to increase in disturbed areas (Fleming and Baldwin 2008). Several of these species, such as E. angustifolium, C. arundinacea, L. pilosa and F. ovina, are adapted to quickly colonize newly disturbed areas, which could explain their dominance. Several of other abundant vascular plants were also pioneering species, such as all the tree seedlings (Betula spp., P. tremula, P. sylvestris and Salix spp.) and the shrub R. idaeus. Most of the bryophytes (Polytrichum spp. and C. purpureus) are also pioneering opportunists, with a rapid increase in biomass immediately after fire (Schimmel and Granström 1996; Fleming and Baldwin 2008). A few late-successional species were also commonly occurring, such as Vaccinium spp. and the graminoid D. flexuosa, even though a short time had passed since the fire. Vaccinium spp. and D. flexuosa are some of the most commonly occurring species in the boreal forest, and have adapted to fire by locating seeds and germinating structures in the lower parts of the mor layer and upper parts of the mineral soil, which protects them from heat penetration (Schimmel and Granström 1996). This could explain their rapid recovery, indicating that even though the wildfire in the Gärsjön catchment area was of high severity, some buds and rhizomes must have survived heat penetration into the ground. The total of 36 species found in the fire area is rather low, and can be compared with 61 ground species three years post-fire in a boreal forest in Estonia (Parro et al. 2009) and 74 species during the first years post-fire in North America (Day, Carrière and Baltzer 2017). Although vegetation inventories using the point-intercept method does not result in a total inventory of the understory vegetation, the low number of recorded species indicates that many species have not yet successfully been re-established in the fire area, at least not in a frequency high enough to be detected in this study. The NMDS analysis, which was performed to see how species composition could be explained by different environmental variables, revealed that remaining humus depth and site index (Fig. 2) could significantly explain the species composition three years post-fire. Schimmel and Granström (1996) have shown that the most important variable affecting the survival and regeneration of species in boreal forest in connection to fire, is the depth of burn. This coincides with the correlation between post-fire humus depth and species composition, since the remaining humus depth and fire severity was shown to be strongly negatively correlated. However, the result shows that the effects of fire, in terms of remaining humus depth, were not the sole important variable explaining species composition. Previous studies focusing on low to moderate fires in boreal forests have showed similar results, where post-fire species assembly mainly was a factor of pre-fire habitat characteristics such as climatic variables, basal area (Boiffin et al. 2015), and pre-fire forest type and soil-moisture holding capacities (Day, Carrière and Baltzer 2017). The result of the NMDS analysis indicates that pre-fire site characteristics, in terms of site index, is important when describing post-fire understory composition also after high severity fires.

4.2 Understory regeneration Biomass accumulation of the understory was in this study measured as LAI and NDVI, which have shown to be positively correlated with gross primary productivity and photosynthetic processes such as respiration and transpiration (Law and Waring 1994; Gower et al. 2001;

11

Lindroth et al. 2008; Lukes et al. 2016). The multiple linear regression analyses, which were performed in order to see the connection between vegetation regeneration and the variables used, revealed that site index was the sole predictor to explain the variation in LAI three years post-fire, where the results show that understory regeneration increases with an increased site productivity. A higher availability of nutrients creates better regeneration conditions both for species that have survived the fire, but also for pioneering species that establish in the fire area (Boiffin, Aubin and Munson 2015), leading to increased values of LAI in sites with high pre-fire productivity. The multiple linear regression analysis with NDVI as the response variable, showed, on the other hand, that vegetation recovery in terms of NDVI was affected by an interaction between fire severity on soil and moss and site index. The interaction shows that the severity of the fire has a stronger negative effect on vegetation regeneration in sites with low pre-fire productivity, than in sites with high pre-fire productivity. Changes in nutrient availability in connection to severe fires is complex and occur in various ways, where volatilization, increased erosion, altered decomposition processes, leaching and changed moisture holding capacities are common effects (Neary et al 1999, Certini 2005; Shenoy, Kielland and Johnstone 2013). The effects do however differ, where previous studies have shown that severe fires can both increase and decrease in nutrients, especially nitrogen, which is a limiting variable in the boreal forest (Shenoy, Kielland and Johnstone 2013). The interaction between fire severity on soil and moss and site index could depend on that understory vegetation in sites with a lower pre-fire productivity experience a stronger nutrient limitation, and thus takes longer time to recover and regenerate in comparison to sites with higher pre-fire productivity. If the nutrient availability has decreased as an effect of fire, it may also be the actual cause of the nutrient limitation in the first hand, but also an enhancing effect on an already nutrient limited site. The regression analysis also revealed, as mentioned, that LAI and NDVI are connected to different variables, where NDVI was explained by both fire severity and site index, and LAI was only connected to site index. This aligns with previous studies, which have shown that fire characteristics are not the sole important variables when it comes to post-fire understory regeneration in boreal forests (Boiffin, Aubin and Munson 2014; Day, Carrière and Baltzer 2017). The partly different result could depend on that NDVI is a more sensitive measure when it comes to lower vegetation, and therefore is able to better catch a variation in understory biomass at lower heights, thus providing a greater accuracy (Payero, Neale and Wright 2004). LAI, as measured in this study, only captures the field layer, and not mosses and other low-growth species. The higher R2-value further indicates that NDVI might be a better measure of understory vegetation recovery, which also implies the importance of mosses and low-growing plants in understory regeneration measurements. Although based on fewer samples, NDVI seems to be a better measurement of the early pre-fire vegetation recovery than LAI, both because it to a higher degree was explained by the predictor variables, but also since it better matches the result of the NMDS, since both pre-fire productivity in terms of site index and hummus depth were related to the species composition. The variation in understory regeneration in terms of LAI and NDVI was further partly explained by different species composition. LAI was positively correlated to the occurrence of species such as E. sylvaticum, P. erecta, C. arundinacea and J. conglomeratus, while NDVI were positively associated with a vegetation dominated by C. canescens and C. ovalis. Both LAI and NDVI were used to measure vegetation recovery due to their different abilities in recording vegetation at different heights, and it was unsure which of the measurements that would be more appropriate. The different result between LAI and NDVI further on emphasize their divergence in capturing biomass accumulation in the understory. After an initial reestablishment of species and a post-fire boost in understory biomass, previous studies have shown contradictory results where the accumulation either declines or continues to increase (Wang and Kemball 2005; Mack et al. 2008; Ma et al. 2016). It is

12

therefore important to have in mind that these results are only describing the effects on understory regeneration three years after the fire occurred, and long-term effects on species composition and regeneration dynamics would need further follow up studies.

5 Conclusions Understory species composition was explained by fire characteristics in terms of remaining humus depth, and pre-fire productivity. Vegetation recovery in terms of LAI was related to pre-fire productivity only, while NDVI was influenced by both pre-fire productivity and fire severity on soil and moss. This means that even after a severe, stand-replacing fire, pre-fire conditions still matter. Moreover, the two measurements of vegetation recovery, NDVI and LAI, differed in their sensitivity to detect different plant species. LAI mainly measured the density of the field layer and high values were associated to a high abundance of E. sylvaticum, P. erecta, C. arundinacea and J. conglomeratus. NDVI captured both the ground and field layer, and high values were associated with an abundance of C. canescens and C. ovalis. The two different variables thus show at least partly contrasting results. According to my results, it can be concluded that NDVI is a more appropriate measure of post-fire re-establishment and recovery of understory vegetation in the boreal forest.

13

Acknowledgements I would first like to thank my supervisor, Johan Olofsson, for offering me this exciting project, for his help during the way and for steering me onto the right track when needed. I would also like to thank Joachim Strengbom for introducing me to the project, and Gustaf Granath for his help during the introduction, for answering my questions and helping me out with data. A big thank you to Maria-Theresa Jessen for being an excellent company and field worker, together with Laura van Galen for voluntarily helping us out in the field for two weeks. Also, of course, a big thank you to my family and friends, especially Jimmy Karlsson, for encouraging words during the whole process.

References

Angelstam, P.K. 1998. Maintaining and restoring biodiversity in European boreal forests by developing natural disturbance regimes. Journal of Vegetation Science 9:593-602.

Boiffin, J., Aubin, I., Munson, A.D. 2015. Ecological controls on post-fire vegetation assembly at multiple spatial scales in eastern North American boreal forests. Journal of Vegetation Science 26:360-372.

Bond, W.J., Keeley, J.E. 2005. Fire as a global ‘herbivore’: the ecology and evolution of flammable ecosystems. Ecology and Evolution 20(7).

Jonasson, S. 1988. Evaluation of the point intercept method for the estimation of plant biomass. Oikos 52: 101–106.

Certini, G. 2005. Effects of fire on properties of forest soils: a review. Oecologia 143:1-10. Day, J.N., Carrière, S. Baltzer, J.L. 2017. Annual dynamics and resilience in post-fire boreal

understory vascular plant communities. Forest Ecology and Management 401:264-272.

Fleming, R.L & Baldwin, K.A. 2008. Effects of harvest intensity and aspect on a boreal transition tolerant hardwood forest. I. Initial postharvest understory composition. Canadian Journal of Forest Research 38: 685-697.

Godall, D.W. 1952. Some considerations in the use of point quadrats for the analysis of vegetation. Australian Journal of Scientific Research/Ser B Biological Sciences 1-41.

Gongalsky, K.B., Persson, T. 2013. Recovery of soil microfauna after wildfires in boreal forests. Soil Biology & Biochemistry 57:182-191.

Gower, S.T., Krankina, O., Olson, R.J., Apps, M., Linder, S., Wang, C. 2001. Net primary production and carbon allocation patterns of boreal ecosystems. Ecological Applications 11(5): 1395-1411.

Hedwall, P-O., Brunet, J., Nordin, A., Bergh, J. 2013. Changes in the abundance of keystone forest floor species in response to changes of forest structure. Journal of Vegetation Science 24: 296-306.

Hudson, J. M.G., Henry, G. H. R. 2009. Increased plant biomass in a High Arctic heath community from 1981 to 2008. Ecology 90(10): 2657-2663.

Kasischke, E.S. and Johnstone, J.F. 2005. Variation of postfire organic layer thickness in a black spruce forest complex in interior Alaska and its effects on soil temperature and moisture. Canadian Journal for Research 35: 2164-2177.

Kasischke, E.S., Christensen, N.L., Stocks, B.J. 1995. Fire, global warming and the carbon balance of boreal forests. Ecological Applications 5(2):437-451.

Kayes, L.J., Anderson, P.D., Puettmann, K.J. 2010. Vegetation succession among and within structural layers following wildfire in managed forest. Journal of vegetation science 21:233-247.

Kruskal, J. B. 1964. Nonmetric multidimensional scaling: a numerical method. Psychometrika 29:115–129.

Law, B.E., Waring, R.H. 1994. Remote sensing of leaf area index and radiation intercepted by understory vegetation. Ecological Applications 4(2): 272-279.

Lindroth, A., Lagergren, F., Aurela, M., Bjarnadottir, B., Christensen, T., Dellwik, E., Grelle, A., Ibrom, A., Johansson, T., Lankreijer, H., Launiainen, S., Laurila, T., Mölder, M., Nikinmaa, E., Pilegaard, K., Sigurdsson, B.D., Vesala, T. 2008. Leaf area index is the

14

principal scaling parameter for both gross photosynthesis and ecosystem respiration of Northern deciduous and coniferous forests. Tellus 60B: 129-142.

Lukes, P., Stenberg, P., Möttus, M., Manninen, T., Rautiainen, M., 2016. Multidecadal analysis of forest growth and albedo in boreal Finland. International Journal of Applied Earth Observation and Geoinformation 52: 296-305.

Ma, J., Bu,R., Liu, M.,Chang, Y., Han, F.,Qin, Q., Hu, Y. 2016. Recovery of understory vegetation biomass and biodiversity in burned larch boreal forests in Northeastern China. Scandinavian Journal of Forest Research 31:4, 382-393

Mack, C.M, Treseder, K., Manies, K.L., Harden, J.W., Schuur, E.A.G., Vogel, J.G., Randerson, J.T., Chapin III, F.S. 2008. Recovery of aboveground plant biomass and productivity after fire in mesic and dry black spruce forests of interior Alaska. Ecosystem 11:209-225.

Metzger, C.M.H., Heinichen, J.,Eickenscheidt, T., Drösler, M. 2016. Impact of land-use intensity on the relationships between vegetation indices, photosynthesis and biomass of intensively and extensively managed grassland fens. 2016. Grass and Forage Science 72:50-63.

Myndigheten för samhällsskydd och beredskap (MSB). 2015. Skogsbranden i Västmanland 2014. Advant Produktionsbyrå. https://www.msb.se/RibData/Filer/pdf/27530.pdf (retrived 2017-10-10).

Neary, D.G., Klopatek, C.C., DeBano, L.F., Ffolliott, P.F. 1999. Forest Ecology and Management 122:51-71.

Nilsson, M-C., Wardle, D.A. 2005. Understory vegetation as a forest ecosystem driver: evidence from the northern Swedish boreal forest. Frontiers in Ecology and the Environment 3(8):421-428.

Oksanen, J., Blanchet, F.G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., Minchin, P.R., et al. (2016). Vegan: Community Ecology Package. R package version 2.4-0. http://CRAN.R-project.org/package=vegan

Parro, K., Köster, K., Jögiste, K., Vodde, F. 2009. Vegetation Dynamics in a Fire Damaged Forest Area: the Response of Major Ground Vegetation Species. Baltic Forestry 15(2):206-215.

Payero, J.O., Neale, C.M.U., Wright, J.L. 2004. Comparison of eleven vegetation indices for estimating plant height of alfalfa and grass. Applied Engineering in Agriculture 20(3): 385-393.

RStudio Team (2015). RStudio: Integrated Development for R. RStudio. Inc., Boston, MA. http://www.rstudio.com/

Ryan, K.C. 2002. Dynamic interactions between forest structure and fire behavior in boreal ecosystems. Silva Fennica 36(1):13-39.

Schimmel, J., Granström, A. 1996. Fire severity and vegetation response in the boreal Swedish forest. Ecology 77(5):1436-1450.

Shenoy, A., Kielland, K., Johnstone, J.F. 2013. Effects of fire severity on plant nutrient uptake reinforce alternate pathways of succession in boreal forests. Plant Ecology 214:587-596.

Smithwick, E.A.H., Turner, M.G., Mack, M.C., Chapin, S. 2005. Postfire Soil N Cycling in Northern Conifer Forests Affected by Severe, Stand-Replacing Wildfires. Ecosystems 8:163-181.

Song, C. 2012. Optical remote sensing of forest leaf area index and biomass. Progress in Physical Geography 37(1):98-113.

Stenberg, P., Rautiainen, M., Manninen, T., Voipio, P. & Smolander, H. 2004. Reduced simple ratio better than NDVI for estimating LAI in Finnish pine and spruce stands. Silva Fennica 38(1): 3–14.

Tagesson, T., Eklundh, L., Lindroth, A., 2009. Applicability of leaf area index products for boreal regions in Sweden. International Journal of Remote Sensing 30(21): 5619-5632.

Van Wagner, C.E. 1983. Fire behavior in northern conifer forests and shrublands. In: Wein, R.W., MacLean, D.A. (Eds.), The Role of Fire in Northern Circumpolar Ecosystems. John Wiley & Sons, New York, pp. 65-80.

Västmanland County Administrative Board, 2015 a. Beslut: Bildande av naturreservatet

15

Hälleskogsbrännan i Västmanlands län [Decision: The establishment of Hälleskogsbrännan nature reserve in the county of Västmanland]. Dnr [511-4243-14]. http://www.lansstyrelsen.se/vastmanland/SiteCollectionDocuments/Sv/om-lansstyrelsen/kungorelser-och-delgivningar/Nya%20naturreservat/Hälleskogsbrännan%20Beslut.pdf (retrieved 2017-09-15).

Västmanland County Administrative Board, 2015 b. Skogsbranden i Västmanland 2014 [The forest fire in Västmanland 2014]. JG Media 21. http://www.lansstyrelsen.se/vastmanland/SiteCollectionDocuments/Sv/manniska-och-samhalle/krisberedskap/Skogsbranden/Dokumentation-Skogsbranden-2014.pdf (retrieved 2017-09-15).

Wang, Q., Adiku, S., Tenhunen, J., Granier, A. 2004. On the relationship of NDVI with leaf area index in a deciduous forest site. Remote sensing of Environment 94(2005): 244-255

Wang, K.K., Kemball, K.J. 2005. Effects of fire severity on early development of understory vegetation. Canadian Journal of Forest Research 35:254-262

Wikars, L-O., Schimmel, J. 2001. Immediate effects of fire-severity on soil invertebrates in cut and uncut pine forests. Forest Ecology and Management 141:189-200.

Supplementary information Table S1: Descriptive statistics of field data. LAI (n=55) and NDVI (n=27).

Variable Median IQR Min Max

LAI (m2 m-2) 0.37 0.28 0.12 1.76 Stand age (yr) 57 74 0.00 130.00 Site index (Pinus sylvestris L.) 22 4.50 16.00 25.00 Fire severity, soil and moss (0-3) 2.73 0.31 2.28 3.00 Fire severity, shrubs (0-3) 2.95 0.10 2.54 3.00 Humus depth (cm) 0.63 1.90 0.00 12.15

NDVI 0.69 0.12 0.31 0.84 Stand age (yr) 52 49 0.00 128.00

Site index (Pinus sylvestris L.) 22 6 0.00 25.00 Fire severity, soil and moss (0-3) 2.73 0.37 2.16 3.00 Fire severity, shrubs (0-3) 2.95 0.15 2.55 3.00 Humus depth (cm) 0.85 2.04 0.00 5.80

IQR: interquartile range.

Dept. of Ecology and Environmental Science (EMG)

S-901 87 Umeå, Sweden

Telephone +46 90 786 50 00

Text telephone +46 90 786 59 00

www.umu.se