The effects of cutting and burning on the regeneration of heathland vegetation in Oldebroek, the Netherlands

The effects of cutting and burning on the

regeneration of heathland vegetation in

Oldebroek, the Netherlands

Name: Martijn Verhoog Supervisor: Dr. A. Tietema Date: July 2, 2016 Bachelor project Earth Science Future Planet Studies

2 Abstract Due to altered land use and changing environmental conditions, the amount of heathlands has steadily decreased since the nineteenth century. Without active management, the remaining heathlands shift to grass- and tree dominated ecosystems. In this study the effects of prescribed burning and cutting on the regeneration and colonization of dry heathland vegetation were researched in the area of Oldebroek, the Netherlands. The impact of these management practices was assessed by determining the average species coverage at two months, 3 years and 7 years after burning, and at 3 years and 7 years after cutting. Furthermore, the NDVI was measured in order to gain insights into the photosynthetic activity after burning and cutting. The vegetation survey showed a steep increase of Calluna vulgaris during the initial three years after burning and a stagnation of the regeneration during the following four years. For cut plots the regeneration of Calluna vulgaris was less pronounced. This difference in regeneration can be largely explained by the differing vegetation age. Keywords: Heathland, regeneration, colonization, burning, cutting, management, vegetation age Samenvatting

Door hervormd landgebruik en veranderende milieuomstandigheden neemt de hoeveelheid heidegebied sinds de negentiende eeuw gestaag af. Zonder actief heidebeheer veranderd heide in een ecosysteem gedomineerd door grassen en bomen. In dit onderzoek zijn de effecten van branden en maaien op de regeneratie en kolonisatie van heidevegetatie onderzocht in de omgeving van Oldebroek, Nederland. De impact van deze beheersmaatregelen is geanalyseerd door de gemiddelde begroeiing van vegetatiesoorten op twee maanden, drie jaar en zeven jaar na branden, en op drie jaar en zeven jaar na maaien te bepalen. Daarnaast is de NDVI gemeten in de gebrande en gemaaide plots, om inzichten te verwerven in de fotosyntheseactiviteit. De vegetatiemeting liet een sterke toename van Calluna vulgaris zien in de eerste drie jaar na branden, welke in de volgende vier jaar stagneerde. Bij maaien was de regeneratie van Calluna vulgaris minder uitgesproken. Het verschil in regeneratie tussen branden en maaien kan grotendeels worden verklaard door het verschil in vegetatieleeftijd tussen de gebrande en gemaaide plots. Trefwoorden: Heide, regeneratie, kolonisatie, branden, maaien, beheer, vegetatie leeftijd

3

Contents

Introduction ... 4 Heathland management ... 4 Cutting ... 5 Processes involved in cutting ... 5 Effects on nutrient availability ... 5 Burning ... 5 Current use of burning and processes involved ... 6 Effects on nutrient availability and pH ... 6 Effects on vegetation ... 7 Constraints on colonization ... 7 Research description ... 7 Research question and sub-questions ... 7 Hypotheses ... 8 Methodology ... 9 Area description ... 9 Methods of analysis ... 9 Vegetation coverage maps ... 9 NDVI ... 11 Statistical analysis ... 11 Results ... 12 Regeneration trends ... 12 Analysis of variance ... 13 NDVI analysis ... 13 Discussion ... 14 The role of age in regeneration ... 14 Hypotheses ... 14 Future methodological improvements ... 15 Suggestions for further research ... 16 Conclusion ... 16 Acknowledgements ... 17 References ... 18 Appendices ... 20 Appendix A: Vegetation maps ... 20 Appendix B: Complete conditional species coverage table ... 35 Appendix C: ANOVA tables and Multiple comparison figures ... 36 Appendix D: Pearson correlation ... 37 Appendix E: Pictures of the burnt plots ... 38

4

Introduction

Human influence has been determinative for the establishment of heathlands, a landform that originally formed from large scale deforestation and possibly also from peat extraction (Diemont 1996). After their formation, heathlands remained in their characteristic state due to agricultural exploitation, as the land was used for cattle grazing and the collection of plaggen. These were slabs of soil that were cut out and put in stables for the collection of manure, which could later be applied to the field as fertilizer (Stuijfzand et al. 2004). The removal of vegetation by grazing and the removal of soil for plaggen caused nutrient-poor conditions, characteristic of heathlands. Since the nineteenth century the amount of heathland in the Netherlands has decreased significantly, as can be seen in figure 1. Heathlands have been increasingly transformed into arable land, which increases the pressure on the remaining heathlands. During the 1960s the conversion of heathlands into arable land was prohibited by law (Diemont 1996).

Figure 1: The decrease of heathlands since nineteenth century (Diemont 1996).

According to Gimingham (1992) it is difficult to assign a precise definition to heathlands, as it relates more to a characteristic type of landscape than to specific flora and fauna. Heathlands vary greatly due to differing conditions and environmental gradients, and are by no means all alike. However, it is possible to distinguish different types of heathland. The most important categories are dry and wet heaths. Both types of heaths are poor in nutrients and acidic, but in dry heaths the soil is well drained resulting in dryer conditions. Contrary to this, wet heaths occur in places where drainage is impeded and peaty humus accumulates at the surface. In this study however, only dry heath systems were considered. As was stated before, up until the twentieth century heathlands played an active role in agriculture and its management was part of farm operations. However, with the introduction of chemical fertilizers the need for plaggen decreased which led to a change in land use (Oosterbaan et al. 2006). During the past decades the increased nitrogen and sulfur deposition from the atmosphere coupled with a reduced influence of base-rich groundwater has led to widespread soil acidification, while an increased nutrient deposition and mineralization has led to increased nutrient availability. The combination of these factors is detrimental for the survival of many characteristic heathland species. Without active management, heathlands shift to grass- and tree dominated ecosystems.

Heathland management

In Dutch heathlands a number of management practices are used in order to maintain the characteristic heathland vegetation. These are: sodding, cutting, chopping, sweeping, grazing, burning, removing woody storage and liming. The choice of management practice is dependent on soil conditions, the type of heather

5 and the severity and type of degradation in heathlands, and often a combination of different management practices is most appropriate (Oosterbaan et al. 2006). For this study however, only cutting and burning are relevant. In this section a literature based overview is given of both management practices and their influence on soil processes and vegetation. Even though all management practices influence faunal populations as well which in turn can influence the regeneration of heathlands, these will not be taken into account in the current study.

Cutting

With cutting of heathland vegetation the aboveground biomass is cut and removed completely. By doing this, the available aboveground nutrients are removed and the formation of a litter layer is impaired. In the Netherlands, cutting of Calluna vulgaris and Erica tetralix generally happens once every ten years. The age of the plant is a determinative factor for the regenerative capacity of Calluna vulgaris. When the plant is younger than ten years it regenerates well, but after ten years the regenerative capacity decreases significantly. Erica tetralix on the other hand regenerates well until the age of twenty years (Oosterbaan et al. 2006). Processes involved in cutting The effects of heathland management practices are very site specific. According to Oosterbaan et al. (2006) cutting generally has a negative effect on heathland vegetation, because the practice leads to a homogenous, low and dense vegetation structure with a lack of open sandy patches. The absence of these open patches negatively influences the species richness of a heathland. Furthermore, moss and lichen species are damaged by cutting. The heavy machinery used for cutting decreases the local relief, which has a negative effect on faunal species such as ants and reptiles. Huston (1994) states that the regeneration after disturbances is heavily influenced by the subsequent reduction in competition after the disturbance. According to Chytrý et al. (2001), a reduction in competition, especially for light, increases the chances for new species to colonize a disturbed patch. They conclude however that only the complete removal of biomass, resulting in bare soil, increases species richness. This is for instance the case with sodding. However, since cutting only removes the aboveground canopy, species present at the time of the disturbance will regrow and few new species will colonize the patch. A study by Sedláková & Chytrý (1999) showed that cutting resulted in the establishment, and if already present before cutting the increase, of grass species in the first few years after cutting. Calluna vulgaris however is a superior competitor for light and thus was able to regenerate slowly and after a few years become dominant again. Effects on nutrient availability As was stated before, the accumulated aboveground nutrients are taken out of the system with the removal of the aboveground biomass. According to Power et al. (2001), the amount of removed nutrients is ultimately dependent on the intensity of cutting, as a low intensity cut removes less nutrients than a high intensity cut. Härdtle et al. (2006) concluded that cutting has a relatively high efficiency in removing nutrients when compared to other management practices. In their study cutting removed 5, 8.2, 36.5, 7 and 14 years’ worth of nutrient accumulation of N, Ca, K, Mg and P respectively. Especially the removal of Ca, K, Mg and P was high when compared to burning.

Burning

European heathlands have a long history of prescribed burning as a way of managing heathlands. The practice has been used since around 1800, and possibly earlier. The main goal of burning heathlands was not to maintain the characteristic species present in heathland ecosystems, as it is now, but it was used as a means for removing the woody, less nutritious and poorly palpable vegetation in order to maintain the most optimal

forage for grazing animals (Mallik & Gimingham 1983).

In the Netherlands controlled burning was applied as a means for rejuvenating heathlands and removing woody shrubs and trees. Burning was applied up until the 1980s, when it was found that the practice was

6 not as efficient for the removal of nutrients as sodding, among others, and that burning had a negative effect on many faunal species. It was not until recent years that prescribed burning received attention again as a successful management tool (Bobbink et al. 2009). Current use of burning and processes involved Since the 1990’s burning has been gaining popularity again in heathland management due to positive results obtained by forestry managers. Prescribed periodic burning of heathlands temporarily removes a large part of the aboveground biomass, resulting in shifted competition between species. These altered circumstances can have a positive influence on the reintroduction or resettlement of species (Bobbink et al. 2009). Whether the outcome of burning is positive or negative is influenced by a range of factors, many of which are

interconnected through feedback mechanisms.

First of all, vegetation age is important. Watt (1955) distinguishes four phases in the development of Calluna vulgaris: the pioneer, building, mature and degenerate phase. Gimingham (1972) estimates these phases at 2 to 6 years of age (pioneer), 10 to 15 years of age (building), 20 to 25 years of age (mature) and beyond (degenerate). A managed Calluna heathland is often subjected to burning in the building phase and it very rarely reaches the late mature or degenerate phase. Heathland vegetation in the building phase is able to regenerate well after burning, because the plant still has twigs on which buds can grow (Bobbink et al. 2009). After this phase the twigs become woody biomass and the regenerative capacity decreases

significantly (Oosterbaan et al. 2006). Furthermore, fire temperature is a determining factor in the outcome of burning. The flame temperature is dependent on the season, humidity, wind speed and available fuel (in the form of biomass) and burning technique. Burning in summer removes much of the litter- and humus layers and woody biomass (Diemont 1996). When applied in winter however, temperatures will remain relatively low due to the moisture in the soil and biomass. With winter burning the litter- and humus layers are mostly untouched (Bobbink et al. 2009). Heathland vegetation shows the highest chances of survival when burning is applied in January or February. The type of fuel is determinant for the temperature, as woody biomass often causes higher temperatures. Finally, the burning technique determines the temperature as well. Two distinct techniques are possible: back-fire and head-fire. With back-fire the fire travels against the wind, which causes a slow progression of the fire front. This results in higher temperatures, with more biomass being burnt. In the case of head-fire, the fire travels with the wind. This technique causes a much shorter exposure to fire and less biomass to be burnt, resulting in lower temperatures (Bobbink et al. 2009). Effects on nutrient availability and pH As was stated before, the effects of fire on heathland ecosystems are difficult to generalize due to the fact that the processes involved are influenced by many factors. This is also the case for the change in nutrient balance after burning. A study by Chapman (1967) concluded that fire releases up to 95% of the available nitrogen from the system, and the release of sodium, potassium, calcium, magnesium and phosphorus were between 20 and 30%. This nutrient budget is far from complete however, as it does not include losses from leaching or additions from precipitation yet. The study by Diemont (1996) elaborates on these results by considering the nutrient losses if the litter and humus layers are taken into account as well. The resulting nutrient losses were much higher than the study by Chapman (1967) suggested (only nitrogen, potassium and phosphorus were measured). Diemont (1996) attributes these differences to the added release of nutrients from the combustion of the litter and humus layers. Diemont (1996) also underlines the variability between sites resulting from varying circumstances (the amount of combustible material, the intensity and

frequency of the fire and the differing soil properties).

A study by Mohamed et al. (2007) showed that prescribed burning of Calluna vulgaris dominated heathlands during winter causes a significant increase of ammonium (NH4+) in the O-horizon. This is attributed to the increased mineralization of roots and other organic material left in the soil after burning. He states that the increased ammonium concentration coupled with limited N-uptake by regenerating vegetation can cause increased N leaching. Furthermore, Dorland et al. (2003) state that elevated ammonium

7

levels can obstruct the recolonization of desired heathland species. The research by Mohamed et al. (2007) also showed a temporary increase of plant available nitrate, phosphate and magnesium and no changes in potassium and calcium concentrations. Leaching of nutrients however did increase significantly due to increased porosity of the soil and limited nutrient uptake by plants (Mohamed et al. 2007). According to Mallik & FitzPatrick (1996) this increased soil porosity is a result of the increased soil temperature and nutrient availability and decreased soil acidity caused by burning.

As with changes in the nutrient balance, the effect burning has on soil pH is highly dependent on the different factors. According to Allen (1964) burning results in an increase of soil pH, while the study by Mohamed et al. (2007) did not show any significant changes in soil pH. Increased soil pH could be the result of the deposition of basic ash (Forgeard 1990), while no significant changes in pH could be the result of increased nitrification after burning (Brady & Weil 1996). Effects on vegetation As was stated before, the age and composition of heathland vegetation prior to burning is determinative for the regeneration after burning. Older Calluna vulgaris is less able to regenerate after burning compared to younger Calluna. Often the species composition after regeneration or recolonization does not differ much from the preexisting vegetation. Many grasses and mosses are able to profit during early succession from the increased nutrient input and decreased competition for light, but the coverage of these species often decreases again during the following years (Bobbink et al. 2009).

Constraints on colonization

The regeneration and recolonization of heathland vegetation after both burning and cutting is either determined by the vegetative regrowth (regeneration) of remaining vegetation or by germination (colonization) from the seed bank. With both management practices underground biomass is not or barely affected so vegetative regrowth is possible (Bobbink et al. 2009). Recolonization from seeds occurs either from within the community, or by dispersal from neighboring communities (Valbuena & Trabaud 2001). Bakker and Berendse (1999) distinguish two constraints associated with recolonization from seeds after disturbance. Firstly, the composition and quality of the seed bank is important. Bakker and Berendse (1999) state that the seed bank often contains a large share of non-target species (or unwanted species) with a long-term persistence, while target- and endangered species are under-represented (target species being the desired species in a certain type of managed heath, these species vary between different types of heath). Furthermore, the seeds from neighboring communities have a limited dispersal range. The dispersal range is partly determined by the morphological characteristics of the seeds themselves and partly by environmental conditions, and even though seeds have various methods of dispersal (wind, via other organisms etc.) the chances of a seed arriving at a suitable site are decreasing due to heathland fragmentation.

Research description

Research question and sub-questions

In order to study the impacts of cutting and burning on the regeneration and colonization of characteristic heathland vegetation, a comparative study was done at a dry heathland area near Oldebroek, the Netherlands. A chronosequence of cutting and burning was made, by establishing the presence and abundance of heathland species in a field setup. The main research question was formulated as:

What is the impact of prescribed burning on the regeneration and colonization of dry heathland vegetation when compared to cutting, in the study area of Oldebroek, the Netherlands?

8 The sub-questions were formulated as: What is the average coverage of species at two months, 3 years and 7 years after burning and after cutting? To what extent does the regeneration and colonization of dry heathland vegetation differ between prescribed burning and cutting? Does prescribed burning or cutting result in the establishment or regeneration of target species?

Hypotheses

The hypotheses used in this research are based on the available literature and current knowledge on the impact of cutting and burning on the regeneration and colonization of heathlands vegetation. The following hypotheses were formulated: Hypothesis 1: In the short term the decreased competition for sunlight resulting from cutting causes grass species to thrive, but they are outcompeted in the long term by Calluna vulgaris. The first hypothesis is based on the research done by Sedláková & Chytrý (1999). They state that repeated observations in Dutch heathlands have shown that grass species that have survived under the canopy cover of Calluna spread rapidly due to the increased light availability. The grass species were able to become established, or if already present spread, but in the long term Calluna became dominant again as it is a superior competitor for sunlight.

Hypothesis 2: In the long term cutting causes a homogenous, dense vegetation structure that is

limited in species richness.

The second hypothesis is based on Chytrý et al. (2001). A long-term comparison in species richness after cutting, sod-cutting and burning revealed that species richness was increased only if the canopy cover, litter layers, moss mats and dense herbaceous vegetation were removed resulting in exposure of bare patches of soil. When only the aboveground canopy is removed, as is the case with cutting, only vegetative regrowth of species already present occurs resulting in a dense homogenous vegetation structure with limited species richness. Hypothesis 3: In the short term mosses and grasses regenerate quickly after burning due to the input of nutrients and reduced competition, but in the long term are outcompeted by

Calluna vulgaris. The third hypothesis is based on Bobbink et al. (2009). The decrease in competition for light and increase in nutrient availability make for a quick succession of grasses and mosses, but in the long term Calluna can outcompete these species. Hypothesis 4: In the long term prescribed burning results in a species richness comparable to before the treatment, or slightly elevated.

The final hypothesis is based on the research of Chytrý et al. (2001). In this study, burning caused more open sandy patches to form compared to cutting, resulting in the establishment of more species. Due to the canopy closure of Calluna in the long term the total species richness did not increase much and was comparable to the conditions prior to burning.

9

Methodology

Area description

The experimental sites were located on the “Oldebroekse Heide” (52°24'N, 5° 55'E) near the towns of Oldebroek and ‘t Harde, at an altitude of 25 meters. The area, which covers approximately 5500 ha, is situated in a military practice ground for heavy artillery of the Dutch Royal Army. The mean annual temperature is 9.4 °C, the area receives an annual mean precipitation of 1005 mm and the growing season is from April to October (Kröel-Dulay et al. 2015). The dominant vegetation is Calluna vulgaris, which according to Chytrý et al. (2001) is a superior competitor in heathlands due to the fact that it is an evergreen species and because it produces slowly-decomposing litter (Van Meeteren et al. 2008). Its growth rate however is slow. Other species present in the area are Deschamspia flexuosa and Molinia caerulea. The dominant moss types in the area are Hypnum jutlandicum and Dicranum scoparium (Kopittke et al. 2013). The soil is predominantly poor in nutrients, well-drained, sandy and acidic. The abundant soil types are Haplic Podzols and Haplic Arenosols (Kröel-Dulay et al. 2015) (Van Meeteren et al. 2008). The site experiences high nitrogen deposition (30-40 kg N/ha/year) (Emmett et al. 2004). Active management is applied on the heathland by periodic prescribed burning during winter. The area is divided into 52 segments of variable sizes. Every year six or seven of these segments are burnt, and in theory each segment is burnt every 8 years. However, the ideal conditions for burning are very narrow. If burning during the winter is not possible due to poor conditions the segment is skipped and burnt after 16 years. Due to this, each segment is burnt on average once every 13 years. An overview of the segments is given in figure 2. The technique of burning that is applied is head-fire, which results in lower temperatures and a relatively untouched humus layer. The choice of management by burning is twofold. The artillery practice in the field causes around 80 accidental fires per year. Burning is applied to rejuvenate the vegetation, decreasing the chance of small accidental fires becoming large scale fires. Furthermore, burning has little effect on the humus layer and thus facilitates a quick regeneration of vegetation (Jansen 2008). Cutting is applied on a smaller scale.

Methods of analysis

Vegetation coverage maps For the analysis of burnt sites three sites were selected that were burnt in 2016, 2013 and 2009 (appendix E). The locations of these sites within the fieldwork area are shown in figure 2. In each of these sites three plots, measuring 2 by 1 meters each, were selected. The selection of plots within each site was performed on the basis of representativeness of vegetation for the area. Of these plots, 1:10 scale maps were drawn on graph paper of the vegetation present on species level. The type of species and the estimation of coverage (if applicable) were recorded. The sites in which cutting was applied were part of the long-term climate experiment of the University of Amsterdam. This experiment consists of three treatments in triplicates, measuring 20 m2 each (figure 3). For the current research only the control plots were used (Kopittke et al. 2013). The location of the climate experiment and the layout of the site is shown in figure 2. In the control sites cutting was applied to an area measuring 2 by 1 meters in 2009. In 2013 the surface surrounding this 2 m2 _{plot was cut. Figure 3 shows}

position of the plots in the field (the plots marked with “C” indicate the cut plots that will be used in this study) and the layout of each plot. Of the cut plots, similar 1:10 scale maps were constructed.

10 Figure 2: An overview of the research site. Yellow lines indicate the 52 segments on which burning is applied. Top enlargement: the position of burnt sites in the field. Within each highlighted site three plots were selected for the analysis of burning. Bottom enlargement: the position of the climate experiment within the field (Ministry of Defense 2007)(Kopittke et al. 2013). Figure 3: Left: Layout of the sites within the long-term climate experiment. The sites marked with "C" are the control sites that were used in the current study (Kopittke et al. 2013). Right: layout of a control site (Schaap 2015). The hand-drawn maps of the burnt and cut plots were digitized using ArcGIS 10.1, by creating polygon features of the different species present. The following attributes were assigned to the polygon features: type of species 1, percentage coverage 1, type of species 2, percentage cover 2 and extra remarks. The multiple species and percentage attributes were used in order to allow for the possibility of multiple layers of species (for instance a moss species growing under Calluna v.). To overcome the problem of a Calluna v.

11 shrub without moss beneath it and a Calluna v. shrub with moss beneath it having the same shrub density, the Calluna v. and moss coverage were taken with respect to the whole polygon area. As a result of this, some polygon features were assigned species coverages exceeding 100%. This is however not uncommon in vegetation analysis (Peet et al. 1998) (Patterson and Best 1996). The attribute tables were exported to Excel 2013. By using the polygon shape area a total percentage of coverage for each species was calculated for each plot and summarized in a conditional species coverage table. In calculating the total area of each plot, only the surface on which a plant was able to grow was taken into account. This excluded metal frames, pipes, plastic surfaces, measurement devices and areas labeled as ‘plot too small’ (this is elaborated upon in the Discussion section). Furthermore, burnt vegetation and burnt litter were treated as bare cover in the burnt plots. NDVI Apart from the vegetation present, the Normalized Difference Vegetation Index (NDVI) was measured in the cut and burnt plots. NDVI is a direct measurement of the “plant greenness” which can be used to draw conclusions on leaf appearance, photosynthetic activity and mortality (Pedersen et al. 2009) (Gamon et al. 1995). NDVI measurements return a value between -1 and 1, with -1 representing snow and ice, 0 representing bare soil and 1 representing the maximum possible photosynthetic activity.

The NDVI was measured using the SKL925 SpectroSense2+GPS from Skye Instruments fitted with two 4-channel SKR1850A light sensors and a removable cosine correction diffuser. The SpectroSense2+GPS was calibrated to the green, infrared, xanthophyll pigment and yellow wavelengths, making use of calibration information provided by Skye Instruments. The green and yellow wavebands were used in the current research, corresponding to NDVI509 and NDVI570 respectively, and a calibration constant was applied to correct

for the cosine correction diffuser. The following equations were used for calculating NDVI (van Alst 2012): 𝑁𝐷𝑉𝐼_%&' = (𝐶𝐻2.×𝐶𝐻11− 𝑍4×𝐶𝐻1.×𝐶𝐻21) (𝐶𝐻2_.×𝐶𝐻1₁+ 𝑍₄×𝐶𝐻1_.×𝐶𝐻2₁) 𝑁𝐷𝑉𝐼_%7&= (𝐶𝐻2.×𝐶𝐻41− 𝑍9×𝐶𝐻4.×𝐶𝐻21) (𝐶𝐻2_.×𝐶𝐻4₁+ 𝑍₉×𝐶𝐻4_.×𝐶𝐻2₁) With: CH1I = Incident green wavelength channel CH1R = Reflected green wavelength channel CH2I = Incident infrared wavelength channel CH2R = Reflected infrared wavelength channel CH4I = Incident yellow wavelength channel CH4R = Reflected yellow wavelength channel Z1 = Calibration constant for NDVI509 Z2 = Calibration constant for NDVI570 The height of the device was set to 1.8 meters from the vegetation, which corresponds to a measurement surface of 0.5 m2_{, and four measurements were taken per plot (Skye Instruments 2004). Weather conditions} ranged from sunny to moderately cloudy. Statistical analysis The NDVI dataset and the species coverage dataset were imported into Matlab R2014b for further statistical analysis. The NDVI data was analyzed for correlation between NDVI and species present in the plots by calculating Pearson’s linear correlation coefficient and its associated P-values. This was done for both NDVI509

and NDVI570, which were compared to Calluna v., total moss, Hypnum jutlandicum and bare coverages in

each treatment. Furthermore, in order to gain insights into the variance of species coverage between treatments, a balanced one-way ANOVA was applied to the coverages between treatments of Calluna v., Hypnum jutlandicum, Dicranum scoparium and bare. These results were summarized using a pairwise multiple comparison test.

12

Results

Regeneration trends The outcome of the species coverage analysis is summarized in the conditional species coverage table (table 1). An overview of the maps and the complete coverage table is given in Appendices A and B. A number of regeneration patterns can be observed in table 1. First of all, the amount of bare soil in the burnt plots decreases significantly during the first three years (between B_2016 and B_2013), but this decrease stagnates during the following three years. As was stated before, bare cover in the burnt plots consists of any burnt vegetation, burnt litter and bare soil. The amount of bare soil in the cut plots on the other hand shows only a slight decrease between the management application years. Furthermore, In the burnt plots the amount of Calluna v. increases significantly during the initial three Table 1: Conditional species coverage table showing the average species coverage per plot. The species are sorted by species type. Green indicates low cover, yellow/orange indicates middle cover and red indicates high species cover. Notable trends in the table are the increase of Calluna v. with the decrease of bare soil during the initial three years after burning, with the stagnation of this trend in the following three years. Hypnum j. increases steadily over time after burning. With cutting the changes over time are less pronounced, with only a slight increase in Calluna and a slight decrease of bare soil and Hypnum j. B1_ 2009 B2_ 2009 B3_ 2009 B1_ 2013 B2_ 2013 B3_ 2013 B1_ 2016 B2_ 2016 B3_ 2016 C1_ 2009 C2_ 2009 C3_ 2009 C1_ 2013 C2_ 2013 C3_ 2013 Heath Calluna v. 77,5 76,2 78,2 89,2 62,8 83,5 0,9 0,9 1,9 20,1 35,1 10,5 10,4 2,1 11,3 Erica c. 0,7 14,1 Moss Campylopus i. 0,3 3,3 28,0 2,1 Cladonia sp. 0,2 0,4 0,1 0,2 0,7 12,6 2,4 Dicranum s. 2,7 1,7 1,1 0,8 0,8 3,0 0,9 9,2 2,4 11,2 20,2 1,9 10,2 Hypnum j. 78,8 65,0 65,7 37,5 33,1 54,7 23,4 22,7 2,8 7,9 0,2 2,9 16,4 6,9 3,4 Polytrichum 0,1 0,6 1,2 0,6 0,6 5,1 4,4 3,8 0,3 0,8 Trees Betula 0,6 0,1 0,2 Pinus 0,2 0,8 0,6 0,8 0,04 3,0 1,1 0,1 0,1 Grass Poaceae 0,1 0,5 0,1 1,5 1,5 3,7 2,8 1,3 7,9 Other Bare 1,5 14,4 4,3 4,6 14,0 4,4 72,7 74,5 95,1 47,9 12,7 61,1 44,7 83,1 64,5 Roots 1,0 0,3 0,7 0,3 4,2 1,1 5,1 4,7 1,7 _{Total 160,8 159,5 151,5 134,0 125,3 145,2 100 100 100 100 100 100 100 100 100 Legend} 0 20 40 60 80 100 Figure 4: Summary of the regeneration trends described in the conditional species coverage table. It is important to note that these trends are not compiled from data of single plots over time, but they are constructed from average trends of multiple plots. In the figure t=0, t=3 and t=7 correspond to the plots treated in 2016, 2013 and 2009 respectively. 0 20 40 60 80 100 0 1 2 3 4 5 6 7 Co ve r ( % ) Time since application (years)

Regeneration trends

Bare after burning Calluna v. after burning Hypnum j. after burning Bare after cutting Calluna v. after cutting Hypnum j. after cutting

13 years, but this regeneration stagnates in the following three years. In the cut plots on the other hand, the regeneration of Calluna v. occurs more slowly. The moss species Hypnum jutlandicum shows a more gradual increase over time in the burnt plots. This species occurred mostly under the canopy of Calluna vulgaris, but it did not follow the same regeneration pattern. In the cut plots this species shows a slight decrease over time. An overview of the regeneration trends is given in figure 4. The moss species Dicranum scoparium had the highest occurrence in the cut plots, with the highest average coverage occurring in the 2013 cut plots (10.8%, see appendix B). Analysis of variance The ANOVA tests on the average species coverage confirmed some of the trends shown in figure 4. First of all, the multiple comparison test of Calluna vulgaris showed that the mean coverages of the plots burnt in 2009 and 2013 differed significantly from the other treatments at a significance level of 5%. However, they did not differ significantly from each other. The mean of the plot burnt in 2016 fell in the range of the plots cut in 2009 and 2013. For Hypnum jutlandicum the means of plots burnt in 2016 and the means of both cut treatments were in the same range. The means of plots burnt in 2013 and 2016 differed significantly both from the aforementioned plots and from each other. The analysis of Dicranum scoparium showed no significant

differences between any of the treatments.

Finally, the mean bare coverage in the plots burnt in 2009 and 2013 showed significant differences between the plots burnt in 2016 and cut in 2013, but not with the plots cut in 2009. The mean of bare coverage in the plots cut in 2009 fell within the range of all other treatments. A complete overview of the ANOVA tables and multiple compare test graphs is shown in appendix C.

NDVI analysis

The results of the NDVI measurements are summarized in figure 5. Both the NDVI570 and the NDVI509

measurements presented similar trends, however at different NDVI-levels. For both wavelengths the cut plots experienced a relatively large spread of measurements compared to the burnt plots. The plots burnt in 2009 and 2013 exhibited the highest mean NDVI, followed by the plots cut in 2009 and 2013. The plots burnt in 2016 displayed the lowest mean NDVI. The one-way ANOVA test affirmed that these three groups were significantly different from each other.

Figure 5: Boxplots of the NDVI results. NDVI570 and NDVI509 corresponds to the yellow and green wavelengths

respectively. For both wavelengths, the plots burnt in 2009 and 2013 show the highest average NDVI, followed by plots cut in 2009 and 2013. The plots burnt in 2016 show the lowest average NDVI. Contrary to the burnt plots, the cut plots display a relatively large spread in NDVI measurements.

The Pearson linear correlation test revealed a high significant positive correlation between NDVI (both wavelenghts) and the presence of Calluna vulgaris, and a moderately high significant positive correlation between NDVI and the presence of moss species. For Hypnum jutlandicum specifically the positive

14

correlation was slightly lower, but still significant. Bare cover demonstrated a decidedly high significant negative correlation with NDVI (table 2). An overview of the correlation results is given in appendix D. The large spread of NDVI in the cut plots can be largely explained by the correlation results. As was shown in table 1 and appendix A the cut plots have a relatively high variability between bare and vegetated soil. Due to the high positive

and negative correlations of vegetation and bare soil respectively, this variability results in a more pronounced spread of NDVI measurements. The burnt plots on the other hand are densely covered by Calluna v., resulting in a narrow spread of NDVI.

Discussion

The role of age in regeneration The trends in the regeneration of heathland vegetation described in this study suggest that especially in the case of Calluna v. the regeneration occurs much more rapidly in the burnt plots during the initial three years than in the cut plots, suggesting that burning is preferable over cutting. However, this trend can be explained largely by the age of the heathland vegetation at the time when management was applied. As was stated before, the age of heathland vegetation is a determinative factor for its regenerative capacity (Oosterbaan et al. 2006). As prescribed burning is applied on average once every 13 years, the vegetation that is subjected to burning never reaches the mature phase and therefore retains its regenerative capacity well (Watt 1955) (Gimingham 1972). The plots subjected to cutting on the other hand are located in an old community within the fieldwork area (figure 3), with heathland vegetation of at least 27 years old (Kopittke et al. 2013). This means that the Calluna v. in the cut plots has reached the degenerate phase in which much of the biomass has become woody and the vegetation has lost its regenerative capacity considerably. For this reason, any conclusion on the comparison between regeneration of heathland vegetation after cutting and after burning is heavily biased in the current study. Hypotheses As was stated in the first hypothesis a reduction in competition for light can cause grass species to thrive during the initial years after cutting is applied, but in the long term Calluna v. becomes dominant again. The high occurrence of bare soil in the cut plots, and thus the reduction in competition for light, can be linked to the age of the Calluna v. at the time of cutting. Because Calluna v. in the cut plots is unable to regenerate well, bare spots remain in these plots. The large spread of NDVI measurements in the cut plots affirm this observation. The results obtained in this study suggest the confirmation of the first hypothesis, as the amount of Poaceae in the cut plots reduces over time while Calluna v. increases. However, the coverage of Poaceae is low in all cut plots and the regeneration of Calluna v. is slow so any definitive confirmation of the hypothesis is inaccurate. Another noteworthy finding is the higher occurrence of Dicranum scoparium in the cut plots, which can also be linked to the abundance of bare patches. This moss species is very drought tolerant, and can thus thrive in the drier conditions associated with bare soil (Klepper 1963). The second hypothesis states that cutting ultimately results in a dense vegetation structure that is limited in species richness. Currently the species richness does not appear to be limited in the cut plots, as a multitude of moss species inhibit the plots. Due to the slow regeneration of Calluna v. however, no definitive conclusions can be made on the hypothesis. However, the NDVI results suggest a dense vegetation structure of predominantly Calluna v. in the burnt plots, as the narrow spread of NDVI corresponds to the high

vegetation cover found in the plots.

NDVI Calluna Moss Hypnum Bare

NDVI509 1 0.89 0.74 0.64 -0.93

NDVI570 1 0.91 0.75 0.66 -0.94 Table 2: Overview of correlation coefficients. Calluna v. showed a high positive correlation at both NDVI wavelengths. The total moss coverage showed a relatively high positive correlation with NDVI, while Hypnum j. showed a slightly lower positive correlation. Bare soil showed a high negative correlation with NDVI. All correlation coefficients are significant. A complete overview of correlation coefficients and associated P-values is given in appendix D.

15

Apart from the vegetation age, the fire temperature is an important factor in post-burning regeneration of heathland vegetation. According to Bobbink et al. (2009), burning during winter and employing the technique of head-fire causes lower flame-temperatures which in turn results in a largely untouched litter and humus layer. This was partly observed in the plots burnt in 2016. A significant share of the litter was burnt but unburnt segments of Hypnum j. and Dicranum s. did remain after burning suggesting relatively low temperatures. All other aboveground vegetation however was affected by burning. As was stated in the third hypothesis, the reduction in competition for light and the increased nutrient deposition after burning may cause an increase in the establishment of grasses in the first few years after management is applied. Even though Poaceae coverage in the B_2016 plots is very low, grass species were not found in the 2013 and 2009 burnt plots. These results suggest the confirmation of the hypothesis that the quickly regenerating Calluna v. outcompetes the grass species. However, because grass species are present mostly in the early years of regeneration and the time between the first and second measurement is three years in the current research, samples of intermediate years should be taken in order to obtain a more complete picture of early succession after burning. The final hypothesis states that the species richness after burning in the long term is comparable to the species richness prior to the treatment. Because the burning strategy at the study site is arranged as an 8-year cycle, conditions prior to burning were not measured. However, the species richness between the 2013 and 2009 burnt plots was comparable, which suggests a confirmation of this hypothesis. The occurrence of Erica cinerea in the 2013 burnt plots is noteworthy however. According to Gimingham (1949) this species is normally found in drier heaths, but during the early regeneration stages of Calluna dominated heaths after burning the two species can be temporarily codominant. The findings of the current research support this claim. Future methodological improvements In the current experimental design and methodology a number of improvements for future research can be pointed out. First of all, the picking of burnt plots was done on the basis of representativeness of the general area. A certain degree of subjectivity is associated with this method. In an ideal situation general characteristics of each burnt site would be selected on the basis of which a randomized selection method could be applied. However, the research site of the current study is located on property of a stakeholder that grants limited access to both the area and to information about the area. These factors combined with relative time constraints led to the subjective method for picking burnt plots. Secondly, some of the cut plots were set out too small during a previous research, measuring for instance 1.8 by 1 meters instead of 2 by 1 meters. In the vegetation maps these sections were mapped as ‘Plot too small’, and this area was not taken into account in the final analysis. The effects of smaller plots on the outcome of the current research are estimated to be limited as the species coverage was calculated relative to the area excluding the ‘Plot too small’ sections, but a smaller sample scale may result in a less representative sample. Previous research by Chytrý et al. (2001) concluded after using a multitude of sample scales (25x25 cm, 1x1 m and 3x3 m) that the scale at which vegetation analysis is performed may yield divergent results. Apart from a smaller sample size, the cut plots also experienced more pronounced edge effects when compared to the burnt plots. In some cases the vegetation of neighboring uncut plots (indicated in figure 3 as the ‘not disturbed’ section) grew into the cut plots, which Figure 6: An example of edge effects in a cut plot. Some vegetation grew into the top of the cut plot from a neighboring uncut plot, interfering with vegetation regeneration.

16 may have interfered with the vegetation regeneration. An example of this is shown in figure 6. In the case of the burnt plots on the other hand the whole field received the same treatment, limiting any edge effects. The results of both the NDVI and vegetation coverage maps of cut plots may therefore be slightly biased. Finally, it is important to note that for any statistical testing performed in the current study the sample sizes were quite small. As a result of this, any statistically significant results may prove insignificant at a larger sample size. Suggestions for further research For further research into the effects of burning and cutting on the regeneration of heathland vegetation it is essential to generalize all parameters and factors influencing the vegetation. In the current research any comparison on the regeneration of Calluna v. after burning and after cutting was biased due to the differences in vegetation age. In an ideal situation both management practices would be applied to vegetation of the same age, and a chronosequence would be made over a multitude of years. Furthermore, because the research took place in an area that is used for the detonation of ammunition an inquiry into the possible contamination caused by ammunition could be fundamental in gaining insights into the development of vegetation in the current location. Both the variability in soil composition within sites and between sites could have significant effects on the regeneration of heathland vegetation.

Conclusions

Without active management, heathlands shift to grass- and tree dominated ecosystems (Oosterbaan et al. 2006). In this study the regeneration of heathland vegetation after burning and after cutting was researched in Oldebroek, the Netherlands. The regeneration trend of Calluna vulgaris after burning observed in this research exhibited a steep increase in coverage during the first three years after management was applied, after which the regeneration stabilized. The cut plots on the other hand showed a slower regeneration of Calluna v. The difference in regeneration however is not necessarily caused by the differing management types, but it can be largely explained by the difference in vegetation age between the cut and burnt plots. Vegetation age is a determinative factor in the regeneration of heathland vegetation, as heathland vegetation loses its regenerative capacity over time (Bobbink et al. 2009). Because the cut plots were located in a significantly older community within the heath compared to the burnt plots, the regeneration of Calluna v. was limited. The bare spots in both the cut and the burnt plots followed a similar trend. In the burnt plots the bare coverage decreased significantly during the first three years after management application and stagnated during the following four years, while the change of bare coverage in the cut plots was less pronounced. The moss species Hypnum jutlandicum increased gradually in the burnt plots, and decreased slightly in the cut plots over time. The higher occurrence of bare soil in the cut plots was supported by the NDVI measurement, which exhibited a larger spread in measurements compared to the burnt plots. The high positive correlation between NDVI and Calluna v. coverage and the high negative correlation between NDVI and bare coverage supported these findings, as a high variability of bare and vegetated soil within a plot causes a large spread of NDVI. Due to the differences in vegetation age between the cut and burnt plots, both the first and the second hypothesis could not be confirmed or rejected. As was hypothesized, a reduction in competition for light causes grass species to thrive during early succession, but Calluna v. becomes dominant in the long term. In the cut plots a reduction in light was abundant due to the occurrence of bare soil, but the coverage of grass species was low and the regeneration of Calluna v. was slow. Furthermore, the formation of a dense, homogenous vegetation structure after cutting was hypothesized. Due to the slow regeneration a dense homogenous vegetation structure did not form. The third hypothesis stated that the reduction in competition for light and the increased nutrient deposition after burning causes the establishment of grass species during early succession. Grass species were only found directly after burning, but the coverage was very low. These findings suggest a confirmation of this hypothesis, but in order to confirm this hypothesis more data of

17 intermediate years should be taken. The final hypothesis, stating that species richness in the long term is comparable to the species richness before treatment, could not definitively be confirmed or rejected as no measurements were taken prior to burning. However, because the species richness in the 2013 and 2009 cut plots was comparable, the confirmation of this hypothesis is suggested.

Due to the differences in age between the cut and burnt plots in the current study, no definitive conclusions could be drawn on the differences in regeneration of heathland vegetation after burning and after cutting. In future research the factors influencing vegetation regeneration should be generalized in order to gain insights into the regeneration of heathland vegetation. Acknowledgements I would like to extend my gratitude to a number of people who made this research possible. First of all, I would like to thank my supervisor Dr. Albert Tietema for his scientific knowledge and his enthusiasm for the field of profession. Furthermore, I would like to thank Henk-Pieter Sterk and Berend Wijers for providing me with knowledge in the field and in the GIS-studio, and especially Henk-Pieter Sterk for his feedback on several occasions. Furthermore, I would like to extend my sincere gratitude to the army personnel at the Artillerie SchietKamp Olderbroek, for their patience and guidance within the fieldwork area. Finally, I would like to thank Brand Timmer for sharing his extensive knowledge on heathland management and for showing me and Stef Knibbeler around the fieldwork area.

18

References

Allen SE. 1964. Chemical aspects of heather burning. Journal of Applied Ecology 1:347-67. Bakker JP, Berendse F. 1999. Constraints in the restoration of ecological diversity in grassland and heathland communities. Trends in ecology & evolution 14:63-8.

Bobbink R, Weijters MJ, Nijssen M, Vogels J, Haveman R, Kuiters AT. 2009. Branden als EGM-maatregel. Directie Kennis, Ministerie LNV. Brady NC, Weil RR. 1996. The nature and properties of soils. 11th _{ed. London: Prentice-Hall} Chapman SB. 1967. Nutrient budgets for a dry heath ecosystem in the south of England. The Journal of Ecology 55:677-89. Chytrý M, Sedláková I, Tichý L. 2001. Species richness and species turnover in a successional heathland. Applied Vegetation Science 4:89-96. Diemont WH. 1996. Survival of Dutch heathlands. Wageningen: Dissertation. Dorland E, Bobbink R, Messelink JH, Verhoeven JTA. 2003. Soil ammonium accumulation after sod cutting hampers the restoration of degraded wet heathlands. Journal of Applied Ecology 40:804-14. Emmett BA, Beier C, Estiarte M, Tietema A, Kristensen HL, Williams D, Peñuelas J, Schmidt I, Sowerby A. 2004. The response of soil processes to climate change: results from manipulation studies of shrublands across an environmental gradient. Ecosystems 7:625-37. Forgeard F. 1990. Development, growth and species richness on Brittany heathlands after fire. Acta Oecologica 11:191-213. Gamon JA, Field DB, Goulden ML, Griffin KL, Hartley AE, Joel G, Peñuelas J, Valentini R. 1995. Relationships between NDVI, canopy structure, and photosynthesis in three Californian vegetation types. Ecological Applications, 5:28-41. Gimingham CH. 1949. The Effects of Grazing on the Balance Between Erica Cinerea L. and Calluna Vulgaris (L.) Hull. In Upland Heath, and Their Morphological Responses. Journal of Ecology 37:100-19. Gimingham CH. 1972. Ecology of heathlands. London: Chapman Hall xv. Gimingham CH. 1992. The lowland heathland management handbook. 1st _{ed. Peterborough, UK: English} Nature. Härdtle W, Niemeyer M, Niemeyer T, Assmann T, Fottner S. 2006. Can management compensate for atmospheric nutrient deposition in heathland ecosystems?. Journal of Applied Ecology 43:759-69. Huston MA. 1994. Biological diversity. The coexistence of species on changing landscapes. Cambridge: Cambridge University Press. Jansen P. 2008. Beheer van defensie als inspiratiebron. Vakblad Natuur Bos Landschap. Klepper B. 1963. Water relations of Dicranum scoparium. The Bryologist 66:41-54. Kopittke GR, Tietema A, van Loon EE, Kalbitz K. 2013. The age of managed heathland communities: implications for carbon storage?. Plant and soil 369:219-30. Kopittke GR, Van Loon EE, Tietema A, Asscheman D. 2013. Soil respiration on an aging managed heathland: identifying an appropriate empirical model for predictive purposes. Biogeosciences 10: 3007-38. Kröel-Dulay G, Ransijn J, Schmidt IK, Beier C, De Angelis P, de Dato G, Penuelas J. 2015. Increased sensitivity to climate change in disturbed ecosystems. Nature communications 6:1-7.

19 Mallik AU, FitzPatrick EA. 1996. Thin section studies of Calluna heathland soils subject to prescribed burning. Soil Use and Management 12:143-9. Mallik AU, Gimingham CH. 1983. Regeneration of heathland plants following burning. Vegetatio 53:45-58. Mohamed A, Härdtle W, Jirjahn B, Niemeyer T, von Oheimb G. 2007. Effects of prescribed burning on plant available nutrients in dry heathland ecosystems. Plant Ecology 189:279-89. Oosterbaan A, De Jong JJ, van Raffe JK, Van Der Heide CM. 2006. Kosteneffectiviteit van beheer van bos-en natuurterreinen. Een onderzoek naar de verhouding tussen kosten en effecten van verschillende maatregelpakketten voor het beheer van droge heide. Alterra 1401. Patterson MP, Best LB. 1996. Bird Abundance and Nesting Success in Iowa CRP Fields: The Importance of Vegetation Structure and Composition. The American Midland Naturalist 135: 153-67 Pedersen JK, Arndal MF, Schmidt IK. 2009. Above and belowground phenology in a heathland during future climate change. IOP Conference Series: Earth and Environmental Science 6:1-2 Peet RK, Wentworth TR, White PS. 1998. A Flexible, Multipurpose Method for Recording Vegetation Composition and Structure. Castanea 63:262-74. Power SA, Barker CG, Allchin EA, Ashmore MR, Bell JNB. 2001. Habitat management: a tool to modify ecosystem impacts of nitrogen deposition?. The Scientific World Journal 1:714-21. Schaap J. 2015. The influence of climate change on heathland vegetation colonization and regeneration in Oldebroek, the Netherlands. Bachelorsciptie Aardwetenschappen, Universiteit van Amsterdam. Sedláková I, Chytrý M. 1999. Regeneration patterns in a Central European dry heathland: effects of burning, sod-cutting and cutting. Plant Ecology 143:77-87. Skye Instruments 2004. Area of Measurement by Reflected Light Sensors. United Kingdom: Skye Instruments limited. Stuijfzand SC, Turnhout CAM, Esselink H. 2004. Gevolgen van verzuring, vermesting en verdroging en invloed van herstelbeheer op heidefauna: basisdocument. Expertisecentrum LNV. Valbuena L, Trabaud L. 2001. Contribution of the soil seed bank to post-fire recovery of a heathland. Plant Ecology 152:175-83. Van Alst N, Tietema A. 2012. Impact of drought in spring on shoot growth, photosynthetic capacity and light use efficiency of Calluna vulgaris in Dutch heathland. Van Meeteren MJM, Tietema A, Van Loon EE, Verstraten JM. 2008. Microbial dynamics and litter decomposition under a changed climate in a Dutch heathland. Applied Soil Ecology 38:119-27. Watt AS. 1955. Bracken versus heather, a study in plant sociology. The Journal of Ecology 43:490-506.

20 Appendices Appendix A: Vegetation maps

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

Appendix B: Complete conditional species coverage table

Table 3: Complete conditional species coverage table. In this table every measured unit is shown. In table 2 a number of units is summarized as bare (for instance burnt litter and burnt Calluna v.). Table 4: Average species coverage per treatment. The data in this table was used in figure 4 (regeneration trends).

36 Appendix C: ANOVA tables and Multiple comparison figures ANOVA 1: NDVI570, variance between treatments ANOVA 2: NDVI509, variance between treatments ANOVA 3: Calluna v., variance between treatments ANOVA 4: Hypnum j., variance between treatments ANOVA 5: Dicranum s., variance between treatments ANOVA 6: Bare, variance between treatments

37

Appendix D: Pearson correlation

Correlation coefficients using NDVI509 and total moss coverage

NDVI CALLUNA MOSS BARE

ALL NDVI DATA 1 0,861101* 0,715684* -0,90297* AVERAGE NDVI 1 0,891423* 0,740886* -0,93477* B2009 1 0,385154 -0,27786 -0,13446 B2013 1 -0,26567 -0,69354* 0,402074 B2016 1 -0,42218 0,426407 -0,42579 C2009 1 0,767217* 0,77414* -0,79035* C2013 1 0,823472* 0,314112 -0,54156 Correlation coefficients using NDVI509 and Hypnum coverage

NDVI CALLUNA HYPNUM BARE

ALL NDVI DATA 1 0,861101* 0,613986* -0,90297* AVERAGE NDVI 1 0,891423* 0,635607* -0,93477* B2009 1 0,385154 -0,2761 -0,13446 B2013 1 -0,26567 -0,69202* 0,402074 B2016 1 -0,42218 0,424355 -0,42579 C2009 1 0,767217* -0,55772 -0,79035* C2013 1 0,823472* -0,0626 -0,54156 Correlation coefficients using NDVI570 and total moss coverage

NDVI CALLUNA MOSS BARE

ALL NDVI DATA 1 0,879254* 0,721603* -0,90508* AVERAGE NDVI 1 0,910898* 0,747573* -0,93765* B2009 1 0,343142 -0,23405 -0,12696 B2013 1 -0,26703 -0,63705* 0,389023 B2016 1 -0,30727 0,332257 -0,32521 C2009 1 0,75042* 0,755348* -0,76558* C2013 1 0,811918* 0,2903 -0,51983 Correlation coefficients using NDVI570 and Hypnum coverage

NDVI CALLUNA HYPNUM BARE

ALL NDVI DATA 1 0,879254* 0,636312* -0,90508* AVERAGE NDVI 1 0,910898* 0,659213* -0,93765* B2009 1 0,343142 -0,23249 -0,12696 B2013 1 -0,26703 -0,63618* 0,389023 B2016 1 -0,30727 0,316463 -0,32521 C2009 1 0,75042* -0,49463 -0,76558* C2013 1 0,811918* -0,08533 -0,51983 P-values using NDVI509 and total moss coverage

NDVI CALLUNA MOSS BARE

ALL NDVI DATA 1 1,1E-18 1,31E-10 6,09E-23 AVERAGE NDVI 1 8,14E-06 0,001578 3,3E-07

B2009 1 0,216322 0,381882 0,676961 B2013 1 0,403951 0,012372 0,195087 B2016 1 0,171582 0,166884 0,167565 C2009 1 0,003586 0,003123 0,002216 C2013 1 0,000996 0,32007 0,068972 P-values using NDVI509 and Hypnum coverage

NDVI CALLUNA HYPNUM BARE

ALL NDVI DATA 1 1,1E-18 1,82E-07 6,09E-23 AVERAGE NDVI 1 8,14E-06 0,010878 3,3E-07

B2009 1 0,216322 0,385032 0,676961 B2013 1 0,403951 0,012645 0,195087 B2016 1 0,171582 0,169157 0,167565 C2009 1 0,003586 0,059537 0,002216 C2013 1 0,000996 0,846736 0,068972 P-values using NDVI570 and total moss coverage

NDVI CALLUNA MOSS BARE

ALL NDVI DATA 1 2,47E-20 7,78E-11 3,32E-23 AVERAGE NDVI 1 2,36E-06 0,001355 2,48E-07

B2009 1 0,274844 0,464064 0,694172 B2013 1 0,40145 0,025884 0,211348 B2016 1 0,331289 0,291347 0,302329 C2009 1 0,004927 0,0045 0,003703 C2013 1 0,001339 0,360017 0,08322 P-values using NDVI570 and Hypnum coverage

NDVI CALLUNA HYPNUM BARE ALL NDVI DATA 1 2,47E-20 4,65E-08 3,32E-23 AVERAGE NDVI 1 2,36E-06 0,007513 2,48E-07

B2009 1 0,274844 0,467127 0,694172 B2013 1 0,40145 0,026154 0,211348 B2016 1 0,331289 0,316263 0,302329 C2009 1 0,004927 0,1021 0,003703 C2013 1 0,001339 0,792023 0,08322 Note: Any significant correlation at a 5% significance level (P=5%) is marked with an asterisk (*).

38 Appendix E: Pictures of the burnt plots Burnt in 2009: Burnt in 2013: Burnt in 2016: