1
LINKING REWILDING &
GEODIVERSITY
Research Proposal
Lukas Struiksma - 10762035
Supervisor and examiner: dr. K. F. Rijsdijk
Assessor: dr. A. C. Seijmonsbergen
Calibrating and Validating
a Weighted Geodiversity Index
for Assessing Rewilding Potential
Calibrating and Validating
a Weighted Geodiversity Index
for Assessing Rewilding Potential
2
Cover image credits: Edit of Peaco, J. (2016). A wolf chases magpies and ravens from an elk carcass in Yellowstone
National Park in 2016 [Online image]. The New York Times.
3
TABLE OF CONTENTS
Key Project Information ... 4
Summary ... 5
Introduction ... 6
Theoretical Framework ... 6
Research Aim and Questions ... 9
Hypotheses ... 9
Methodology ... 10
Time Schedule ... 14
Scientific Embedding ... 16
Knowledge Utilisation... 16
Equipment ... 16
Budget ... 17
Safety ... 18
References ... 19
4
KEY PROJECT INFORMATION
Applicant information
Name: Lukas Struiksma
E-mail: lpstruiksma@gmail.com Date of birth: 14-09-1996
Institution: University of Amsterdam Applying for: MSc project (5 months)
Research group composition
Name and title Institution Involvement
L. P. Struiksma University of Amsterdam Student responsible for the MSc project
Dr. K. F. Rijsdijk University of Amsterdam Examiner, daily supervisor Dr. A. C. Seijmonsbergen University of Amsterdam Assessor, GIS specialist
5
SUMMARY
Rewilding is a conservation approach with momentum: even though its theories originate barely 20 years ago, it has found worldwide adoption. The restoration of stable trophic systems is a key part within (trophic) rewilding theory, which is accomplished by reintroducing apex consumers into ecosystems where they are underrepresented or have completely disappeared. The geodiversity of a given ecosystem is considered to drive rewilding success in multiple ways: it increases resource availability and enhances predator’s ecological impact through establishing a landscape of fear. This research proposal aims to design a method for investigating the link between geodiversity and rewilding potential. Primarily, it tries to distill key insights, concepts and criteria of rewilding implementation into measurable parameters. These are then calculated and classified during an automated geospatial analysis for 30 research areas in temperate climate zones. The resultant scores are compared to the rewilding scores of 20 research areas, and an optimization algorithm calculates the appropriate weight parameter for each geodiversity component. In this way, the difference in influence the various geodiversity components have on rewilding success can be determined. The resultant weighted geodiversity index is validated by comparing its scores to rewilding scores of the remaining 10 areas. If these scores match closely, the weighted geodiversity index may prove to be a helpful tool to establish the rewilding potential of a given area by using its geodiversity as a proxy.
6
INTRODUCTION
During the past century, both human population and its resource usage has grown significantly. This growth has had major effects on societal and environmental conditions, leading to large-scale loss and degradation of biodiversity and ecosystems (Dirzo & Raven, 2003). With this development, many integral ecological processes halt or slow down, which decreases the ecosystem’s complexity and resilience to further threats (Oliver et al., 2015). This grave prospect has led conservationists and environmental managers worldwide to explore and adopt conservation strategies in order to restore ecosystems to historical benchmarks or stagnate their degradation processes. Conservation efforts have historically experienced varying rates of success, with ineffectively managed areas and continuing species loss as a result (Laurance et al., 2012). Other ecosystems have already breached their tipping points insofar that restoration to historical states is no longer an option,
and fragile, novel ecosystems arise. These are often not well understood and do not benefit significantly from conventional conservation methods (Pettorelli et al., 2018). Thus, a novel approach for ecosystem management is required to restore or establish the vital ecological processes that ensure an ecosystem’s proper functioning and resilience to external disturbances. One such approach that has recently emerged and gained significant momentum is the concept of rewilding. The overarching ecological concept of rewilding can be broadly summarized as“the reorganization of biota and ecosystem processes to set an identified social-ecological system on a preferred trajectory, leading to the self-sustaining provision of ecosystem services with minimal ongoing management” (Petorelli et al., 2018). While this trajectory is initially determined and influenced by human actions, the end goal of rewilding is the gradual reduction of human inputs, leading to a highly resilient and self-sufficient ecosystem.
THEORETICAL FRAMEWORK
Trophic rewilding
However, despite (or perhaps because of) the popularity of rewilding, there is ongoing debate about the required criteria for its successful implementation
(Perino et al., 2019). This lack of consensus has led to the adoption of many practical approaches in order to fulfill the broader ecological concept of rewilding. The most commonly utilized approach of applying the concept of rewilding to ecosystems, and one that stays close to its original definition of “large, strictly protected core reserves, connectivity and keystone species” (Soulé & Noss, 1998), is trophic rewilding. This approach is focused on the reintroduction and protection of species in order to restore stable trophic networks that allow the ecosystem to self-regulate. Key participants within these trophic networks are the so-called ‘apex consumers’, mostly large-bodied carnivores and herbivores. These are some of the most important keystone species, so called because their effects on prey,
resources and competitors can largely shape the functioning of their residing ecosystem.
These alterations often occur in the form of top-down trophic cascades, where the presence or abundance of apex consumers has a large, diversifying and stabilizing influence on some or all trophic levels underneath (Svenning et al., 2016). For example, large herbivores influence taxa abundance and diversity by providing dung, altering the landscape by grazing and dispersing seeds. Large carnivores, on the other hand, create spatiotemporal heterogeneity in the behavior and population sizes of mesopredators and herbivores, or increase scavenging and keep the nutrient cycle running by increasing carrion occurrence (Perino et al., 2019; Svenning et al., 2016).
However, owing to their large body size, metabolic requirements and lengthy reproductive cycle, apex consumers require large roaming ranges and are
7
especially vulnerable to human encroachment (Dirzo et al., 2014). Many original apex consumers have (long ago) disappeared from their respective ecosystems, and their associated contribution to stable trophic networks is no more. In trophic rewilding, these species (or in case of extant species, proxy species that approximate the original species’ behavior) are reintroduced in ecosystems, with their role in trophic cascades minimizing the need for human intervention (Pettorelli et al., 2018).
Measuring rewilding
Rewilding approaches are more oriented towards restoring ecological processes than towards restoring ecosystems to historical benchmarks (Higgs et al., 2014). This means that conventional monitoring methods for conservation success, which often use historical data of the ecosystem or reference ‘ideal’ ecosystems as a template, do not fully reflect rewilding successes. Recent research has started to recognize the importance of ecological processes, and has begun to embed it within conservation monitoring approaches
(Hughes et al., 2016).
A monitoring method explicitly focused on rewilding efforts is the rewilding score proposed by
Torres et al. (2018). It frames the condition of an ecosystem as a function of human inputs and outputs over the ecosystem’s natural functions and ecological integrity. These two dimensions are influenced by a wide range of pressure and state variables, which have been concretized into 18 measurable parameters. In order to gain insight in the state of the ecosystem, local practitioners are asked to rate the ecosystem’s performance for each parameter. If these scores are then entered into the formula proposed in the study, a normalized rewilding score can be calculated. Since this score can change over time as the ecosystem’s vital processes are restored, the rewilding score is not only applicable as a measure of rewilding success, but also as a monitoring approach to rewilding progress.
Landscape of fear
Within the trophic system, apex predators can not only regulate spatiotemporal prey density and
behavior by direct engagement (predation), but also by modifying prey behavior and spread by creating a ‘landscape of fear’. When prey animals have knowledge of the presence of apex predators within their foraging ranges, their foraging behavior will change depending on the perceived risk of predation (Laundré et al., 2010). This often leads to situations where the foraging strategy of prey animals does not revolve around maximizing their nutrient intake (with the associated maximized decrease in vegetation) within areas where they perceive predators might be lurking (Ripple & Beschta, 2007). Prey animals are shown to have strong learning capabilities, and since predator efficiency (kill-to-fail ratio) is typically between 8 and 26%, many animals have first-hand experience with predators and live on to learn from this strong motivator (Laundré et al., 2010).
After the reintroduction of wolves in Yellowstone National Park (USA), elk populations left the now risky large open spaces en masse and sought refuge by foraging near the safer forest edges (Ripple & Beschta, 2007). This indicates that there is a strong connection between the possible establishment of a landscape of fear and physical landscape properties: for example, wide and open areas or valleys between higher vantage points increase the predator’s ability to strike, and prey animals are aware of this.
Geodiversity
Landscape features are more and more included as an important concept within conservation frameworks in the form of geodiversity (Thomas, 2012). Similar to rewilding, it is a concept with various competing definitions based on the circumstances it is utilized in. However, one key definition equals geodiversity to “the natural range (diversity) of geological, geomorphological and soil features” which “includes their assemblages, relationships, properties, interpretations, and systems” (Gray, 2005).
In several studies, the components constituting geodiversity have been found to have beneficial impacts on resource availability such as water, energy and nutrients, which in turn enhance an area’s biodiversity
8
(Thomas, 2012). Thus, geodiversity seems to promote rewilding efforts in at least two different ways: it increases biodiversity by enhancing resource availability, and it increases the opportunity for a landscape of fear to arise, which allows predators to far extend their influence on the trophic system beyond direct predation.
Rewilding projects
Real-world applications of trophic rewilding are found on a global scale, but the most thorough implementation so far has been within Europe. Before human expansion, it consisted of large wild and interconnected areas where apex consumers could freely disperse. However, in the past few centuries, many of these apex consumers, such as aurochs, wolves and lynxes, have gone extinct either entirely or within large parts of their previous ranges. Concurrently, the landscape has fractured significantly, limiting connectivity between dense biodiverse core regions. With a current biodiversity crisis in Europe, it is a major region of interest for rewilding conservationists to apply their approach (Vera, 2009).
One very early example of rewilding occurred in the Oostvaardersplassen nature reserve in Flevoland, the Netherlands. Since the area originated due to the land reclamation projects of the 1960s, there was no historical ecosystem and all ecological processes had to be kickstarted from scratch. Originally a temporary natural reserve due to the high number of (up until then) rare bird and waterfowl species, the area risked encroachment of woody vegetation that would turn the valuable wetland into woodland. To combat this, large herbivores like horses, cattle and deer were introduced to graze the kindling woody vegetation and allow for a more diverse vegetation structure to arise. Initially, this approach was largely successful, with emergent ecological processes that were unforeseen as a result. However, in recent years, overgrazing by large herbivores and geese has drastically decreased
vegetation heterogeneity, which in turn decreased biodiversity in bird, insect and small mammal species
(Lorrimer & Driessen, 2014).
Explaining the failures of the Oostvaardersplassen is not an easy feat, due to the recency of the rewilding paradigm and a relative lack of understanding of many of its processes. However, examining its key theoretical criteria, there are some possible causes. For one, a connection to other nearby nature reserves was foregone after local farmers voiced their opposition, which removed the connectivity of the area with other dense core regions (Kopnina et al., 2009). Secondly, while the area contained apex consumers in the form of large herbivores, they faced no predation from apex predators. While population control resulting from predation is still a hotly debated topic with no clear consensus, the spatiotemporal spread of vegetation associated with their presence is also absent (Vera, 2009). Finally, and importantly for the purposes of this research, the Oostvaardersplassen are not a highly geodiverse area (Llano, 2020), which limits geodiversity’s influence on biodiversity. It is important to realize that many of the key decisions that led to the failures in the Oostvaardersplassen were taken before the framework of rewilding was introduced, and much has since been added to the body of knowledge surrounding its practices.
Despite some rewilding testing grounds experiencing failures, Europe now contains more rewilding projects than ever. One important player within the field is Rewilding Europe, a pan-European organization working together with local partners to run eight main rewilding areas across the continent. In conjunction, it maintains the European Rewilding Network, which is an online network linking both the in-house rewilding projects and external European rewilding projects together, with the aim of sharing knowledge and experience (Schepers & Jepson, 2016).
9
RESEARCH AIM AND QUESTIONS
The process of ecosystem loss and degradation that has accelerated during recent times is unparalleled in modern history. Even under conservative assumptions, species loss due to human action is up to 100 times higher than background extinction rates. Several studies therefore indicate that we are in the midst of a sixth mass extinction event, and that our collective actions are largely to blame (Ceballos et al., 2015). The impacts of these decreases in biodiversity and ecosystem functioning do not limit themselves to the ecosystem, but also negatively affect many of the key ecosystem services humanity takes for granted
(Cardinale et al., 2012). Due to the severity of ecosystem degradation, some ecosystems have crossed over their tipping points and have entered a new semi-stable state, which inhibits the possibilities of restoring them to historical conditions (Pettorelli et al., 2018).
Rewilding offers a novel and needed approach to tackling these issues, but its novelty is simultaneously its pitfall: not much research is available to define its criteria or measure its success. Additionally, the connection between geodiversity and biodiversity has been established (Thomas, 2012), but its relationship to rewilding efforts has not been researched thoroughly. One approach that tries to investigate the relationship between rewilding success and geodiversity is that of
Llano (2020), which compares rewilding scores for three areas with geodiversity scores. This study, however, approaches the construction of a geodiversity index as elemental arithmetic: all geodiversity components are added together, leading to a geodiversity index where all components carry equal weight. This obfuscates the difference in influence some geodiversity components might have.
Therefore, this research proposal will concern the construction of a weighted geodiversity index, where the influence of each geodiversity component is calibrated using rewilding scores following the approach of Torres et al. (2018). Using this method, it aims to provide more insight in the importance of several geodiversity components in driving eventual rewilding successes. If this weighted geodiversity index proves to be a good fit for predicting rewilding success, it may aid future rewilding programs by designating which areas could successfully be rewilded (with little human intervention needed) and which areas would have a more difficult or even impractical rewilding trajectory.
In order to cordon off what answers the study should aim to provide, a central research question is posited:
“To what extent can a weighted geodiversity index help assess the potential of rewilding projects within temperate natural areas?”
The following subquestions are used to unpack the main research question into concrete problems that can be answered following the research plan as proposed within this proposal:
1. “Which landscape features exhibit significant influence on key rewilding criteria?”
2. “How can the diversity of these landscape features accurately be quantified?”
3. “How closely does a Weighted Geodiversity Index (WGDI) correlate with rewilding success?”
HYPOTHESES
Based on the theoretical framework, a few inferences can be made about the relationship between geodiversity and rewilding potential and success. First of all, an increase of geodiversity will likely coincide
with a higher rewilding score. Thomas (2012) has shown through remote sensing studies that geodiversity influences resource availability, a major driver of local biodiversity. Additionally, case studies in Yellowstone
10
National Park have indicated that the physical landscape can influence the landscape of fear, which allows for predators to have more ecological impact and a larger role within the trophic system (Ripple & Beschta, 2007). In Llano (2020), an unweighted geodiversity index correlated with rewilding scores via
Torres et al. (2018), indicating that the two might be linked.
However, this connection might be less strong as geodiversity increases significantly. In Llano (2020), the Oostvaardersplassen and the Swiss National Park had comparable rewilding scores, but the geodiversity of the latter was considerably higher. A possible explanation might be that very high geodiversity starts
to limit ecological processes and actors. For example, an alpine ecosystem with large elevation differences and associated steep slopes likely has a smaller area that is suitable for flora and fauna. Additionally, unsuitable and highly geodiverse areas limit connectivity between cores, since they cannot easily be crossed by species, diminishing their ability to disperse.
Since the research by Llano (2020) only
analyzed three research areas, these assumptions can be investigated in more detail within this research, since it contains 30 areas in total. Additionally, the geodiversity index will be weighted based on rewilding scores from several of the areas, which is likely to increase the relationship between rewilding scores and geodiversity.
METHODOLOGY
The research plan I propose can roughly be divided into three distinct, but sometimes overlapping categories: literature review and synthesis, geospatial analysis, and statistical
weighting and validation. A schematic overview of the proposed workflow can be found under
figure 1, and will be elaborated upon below. The initial phase of the actual research will consist of a thorough literature review. Since the concept of rewilding is relatively new and its definition and criteria are fast-evolving (Perino et al., 2019), it is important to gain a broad and up-to-date understanding of the current state of research. The major focus in this phase lies on two main questions:
“what are the key criteria for instigating and maintaining the necessary ecological processes for rewilding?”, and “how can these criteria possibly be influenced by the landscape?”.
When satisfying and well substantiated answers to the questions are found, the knowledge gained during this phase will be used to synthesize a conceptual framework containing key goals of successful rewilding. These goals will then be deconstructed and concretized in a stepwise manner, in order to retrieve their corresponding measurable parameters. In order to answer the main research question, I also need a metric for rewilding success. Acquisition of rewilding scores will primarily occur through correspondence with the
authors of the original paper that introduced the score
(Torres et al., 2018). Any natural areas with an already calculated rewilding score will be added to the list of 30 research areas, with the selection process for the remaining empty spots elaborated upon further below. In tandem with this literature review and synthesis, the first geospatial analysis of the research will be conducted. To aid the understanding of landscape factors that aid rewilding success, both Yellowstone National Park (USA) and the Lausitz region (Germany) will be analyzed. Both of these areas historically lacked apex predators and their important contribution to stable trophic networks, but have both recently been colonized by wolves (Fechter & Storch, 2014; Ripple & Beschta, 2007). Since the reintroduction of wolves in
Yellowstone National Park saw an increase in both the abundance and height of aspen trees in wolf-affected areas (Ripple & Beschta, 2007) while they scarcely changed in areas uncolonized by wolves, this spatiotemporal analysis will focus on vegetation patterns. A normalized difference vegetation index
11
(NDVI) for these areas will be derived in Google Earth Engine. The NDVI will be calculated for four different points in time: four years before the introduction of wolves, the year in which the wolf was introduced, four years after the fact, and eight years past the introduction of the wolves. These timespans are important: moose populations in Yellowstone National Park have been shown to adapt and stabilize within a couple of years after the sudden introduction of wolves (Berger, 2008), which might manifest itself into differing vegetation patterns. Since this time period overextends most satellite’s orbital lifespan, data for the NDVI calculation will be derived from different sources, such as the Moderate Resolution Imaging Spectroradiometer (MODIS) as used by Potter (2019). In order to exclude changes due to seasonal variability in the NDVI, the input data used for its calculations for the four time periods should ideally be from the same month.
While the NDVI has certain caveats and has a history of being used to derive results for which it is not applicable (Yengoh et al., 2015), it is a sufficient method for the purposes of this research. A detailed quantitative assessment of vegetation at high resolutions is not necessary, since the analysis is focused on broad trends. In order to avoid the influence of cloud cover on NDVI values, satellite data will be chosen from clear days. Finally, to limit the influence of soil moisture content on NDVI fluctuations, satellite data will also be chosen from days with at least one preceding day without precipitation.
The eight resulting NDVI datasets will be exported to ArcGIS, where the NDVI rate of change between the four time periods will be calculated. A high rate of NDVI change in between the time periods is a likely indicator of the establishment of a landscape of fear within the surrounding area, showing that the top-down regulatory function of the local wolf population has started - a major rewilding component (Laundré et al., 2010). In order to couple the detected landscape of fear with attributes of the physical landscape, the areas with high rates of NDVI change will be cross-referenced with a digital elevation model (DEM), lithology, soil, hydrological and land cover maps. The various
landscape features that co-occur with high rates of NDVI change will be noted, and serve as an additional source of information for the construction of the conceptual framework.
With the completion of the conceptual framework, all of the geodiversity components of interest are concretized into a set of measurable parameters. In this way, subquestion 1 can already be answered. This also allows for the selection process for the 30 required research areas. The selection process is dependent on data availability for the required geospatial datasets, existing history of local rewilding efforts, and the availability of already calculated rewilding scores. If there are fewer than 30 areas with existing rewilding scores, I will select the required number of remaining research areas plus five, and the questionnaire used by
Torres et al. (2018) will be sent to representative practitioners for each area. The presence of excess areas allows for the possibility that some practitioners will not respond. Using the answered questionnaires and the formulas used by Torres et al. (2018), the rewilding scores for all remaining areas will be calculated. Since the research proposal specifically focuses on trophic rewilding, these remaining missing areas should ideally contain both large herbivores and apex predators. A helpful tool to find these areas is the list of areas within the European Rewilding Network. Utilizing the search function within their database, areas with active wolf, lynx and bear populations will be selected as research areas.
Datasets necessary for the calculation of all geodiversity components will be searched for and retrieved prior to the main geospatial analysis phase. Special attention will be paid to the consistency of categorical datasets in between different research areas, a sufficient resolution to derive meaningful results (dependent on the geodiversity components chosen) that is similar in between research areas, and their availability within open-access repositories. If resolution differences between datasets for the same component between research areas are significant, the finer datasets will be interpolated to fit the resolution of the coarser datasets.
12
Three research areas that are representative in both the total variety in landscape (elevation differences, vegetation types, etc.) and geographical spread of the collection of 30 research are will be selected from the complete list and loaded into ArcGIS for analysis. These will act as experimental grounds for the geospatial analyses techniques required for the calculations of the chosen geodiversity components, which will be based on the approach used by Llano (2020), with additions that may arise due to the inclusion, exclusion or alteration of certain geodiversity components compared to that study. The research areas will be cordoned off by a shapefile, after which a fishnet grid will be overlaid with a cell size dependent on the resolution of the geospatial datasets to be used, as well as the scale on which the ecological processes derived from the conceptual framework generally take place. With the zonal statistics tool, the geodiversity components derived from the conceptual framework will be calculated, after which their variety or diversity within each cell within the research area shapefile will be determined. Based on the theoretical backing of the conceptual framework, the whole range of values of each geodiversity component within individual cells across the three areas will be divided into four categories (or scores), with 1 expected to have the least beneficial influence and 4 expected to have the most beneficial influence for rewilding purposes. Each cell will subsequently be given this score for that geodiversity component, so all cells eventually contain the same number of individual scores as there are geodiversity components in the conceptual framework.
All steps in this geospatial analysis will be recorded along with the boundaries chosen for each category, since these will need to be automated further along the research process. This initial manual calculation of all scores is important: it allows for some trial and error which is more easily fixable than when it occurs in an automated script, and it allows for the picking of representative and meaningful boundaries for each category. The analysis performed during this section, in conjunction with information from the conceptual framework, form the answer to subquestion 2.
Based on the record of manual calculation and analyses, a flowchart will be constructed to illustrate all the (successful) steps taken and to help with the further automation process (similar to those of Benito Calvo et al. (2009) and Forte et al. (2018)). Using this flowchart, a workflow will be built in ArcGIS ModelBuilder containing all steps necessary to calculate all cell geodiversity scores. The end product of this model run should be the average score for each geodiversity component across all cells within each research area. Finally, this model will be run for each of the remaining 27 research areas, and their average geodiversity scores will be exported into a table.
In the final phases of the research, the data retrieved from the geospatial analyses will be used for statistical weighting and validation. Twenty research areas will be picked at random using the random sample function in Matlab. The average geodiversity scores for these 20 areas will be entered into the following formula in order to calculate each area’s Weighted Geodiversity Index (WGDI):
𝑾𝑮𝑫𝑰 = 𝑾𝟏∗ 𝑮𝑪𝟏+ 𝑾𝟐∗ 𝑮𝑪𝟐+ 𝑾𝟑∗ 𝑮𝑪𝟑 ⋯
Where:
Wx = the unknown weight parameter for geodiversity
component x
GCx = the average score for geodiversity component x
Using a self-constructed Matlab optimization algorithm, the normalized WGDI scores will be compared to the normalized rewilding scores. By minimizing the average discrepancy between the two scores for all of the areas, the best-suited weight parameters for all geodiversity components are calibrated. In a final step, the normalized WGDI with optimized weight parameters will be divided by the non-normalized WGDI. All weight parameters will be adjusted by the resultant ratio, so that normalization of the WGDI and rewilding scores is no longer necessary.
With the weight parameters calculated during the previous phase, the average geodiversity scores for the remaining 10 areas will be entered into the WGDI formula. In order to validate whether or not the WGDI
13
can act as a predictive indicator for rewilding success, the subsequent WGDI scores will be compared to their respective rewilding scores. Finally, the correlation
between the two will be tested using the Pearson correlation coefficient. The resulting correlation coefficient provides the answer to subquestion 3.
Figu re 1: a s ch ematic de p icti on o f th e w or kf lo w as de scr ib ed i n de ta il i n th e s ect io n ab ov e. A ll s te p s ar e co lo r co de d to de p ict th e cate go ry th ey b elo n g t o.
14
TIME SCHEDULE
I expect to need 20 weeks of fulltime labor in order to complete the proposed thesis. Since this entails all of the steps described in the methodology as well as a significant writing period, it is highly important that all phases are planned out meticulously. What follows is a schematic overview of the time schedule under figure 2, with a concise description of the main goals, products and deadlines of each individual step in the schedule.
Figure 2: a schematic overview of the phases and subphases of the thesis process per week. Color codes of the subphases correspond to those in the schematic workflow of figure 1.
Literature review
Deadline: 17/01/2021
Goals/products: a knowledge of the recent state of
research with regards to rewilding theory, trophic systems, landscapes of fear and geodiversity.
Collecting rewilding scores
Deadline: 07/02/2021
Goals/products: a full list of the rewilding scores for 30
research areas, primarily retrieved through correspondence with Torres (2018). If this does not provide enough scores, correspondence to practitioners within chosen research areas will sent out, with a month of time for them to respond.
Calculating NDVI and rates of change
Deadline: 17/01/2021
Goals/products: four NDVI maps over several time
periods for Yellowstone National Park and Lausitz, along with three calculated rate of change maps for both areas.
Investigating landscape and NDVI overlap
Deadline: 24/01/2021
Goals/products: an understanding of the spatial
overlap of key landscape features with areas where NDVI rate of change is high.
week 50 51 52 53 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Literature review
Collecting rewilding scores
Calculating NDVI and rates of change Investigating landscape and NDVI overlap Synthesizing conceptual framework Research area selection
Dataset selection
Manual geodiversity calculation Geodiversity reclassification Automated geodiversity calculation WGDI calibration
WGDI validation Writing concept thesis Revising thesis Holiday period April 2021 Interruptions Preparatory phase Research phase
Weighting and validation phase Writing phase
15
Synthesizing conceptual framework
Deadline: 31/01/2021
Goals/products: a conceptual framework built on key
concepts within rewilding theory, which are concretized into measurable landscape-related parameters.
Research area selection
Deadline: 07/02/2021
Goals/products: a list of 30 (+ 5) research areas, where
datasets are openly accessible and rewilding efforts (or similar) have taken place.
Dataset selection
Deadline: 07/02/2021
Goals/products: a collection of open-access datasets
covering all research areas with sufficient resolution that allow for the calculation of all necessary geodiversity components.
Manual geodiversity calculation
Deadline: 21/02/2021
Goals/products: the average cell value for each
geodiversity component over three representative research areas, as well as a record of the steps taken to calculate these values.
Geodiversity reclassification
Deadline: 21/02/2021
Goals/products: a categorization of all geodiversity
scores per cell into four categories, based on the knowledge from the conceptual framework and the entire range of cell values across the three research areas.
Automated geodiversity calculation
Deadline: 07/03/2021
Goals/products: a model capable of performing all
necessary calculation steps in order to calculate the cell geodiversity scores for the remaining 27 research areas,
as well as a table containing the average cell geodiversity score for each geodiversity component for all areas. WGDI calibration
Deadline: 07/03/2021
Goals/products: calibrated weight parameters for each
geodiversity component within the WGDI based on optimizing the WGDI of all areas with regards to the rewilding scores of these areas.
WGDI validation
Deadline: 07/03/2021
Goals/products: Pearson’s correlation coefficient for
the relation between the WGDI of the remaining 10 research areas and their rewilding scores, using the pre-calibrated weight parameters.
Writing concept thesis
Deadline: 25/04/2021
Goals/products: a full concept version of the thesis, that
can be sent out to individuals related to the research projects in order to receive feedback.
Revising thesis
Deadline: 09/05/2021
Goals/products: the final version of the thesis,
incorporating all feedback received on the concept version.
Interruption: holiday period
Period: 28/12/2020 to 10/01/2021
The period around Christmas and New Year’s Eve is a holiday period, so my examiner and assessor are likely not reachable. This period is initially counted as an interruption period, but some work may still be done, leading to me having more flexibility in my schedule for later phases.
16
SCIENTIFIC EMBEDDING
This research will primarily be conducted within the Institute for Biodiversity and Ecosystem Dynamics (IBED) within the University of Amsterdam. Additionally, it is planned to communicate and collaborate with dr. P. Cornelissen, senior ecology consultant at Staatsbosbeheer (State Forestry Service) and researcher at the Oostvaardersplassen. The
research also coincides with a larger research project of Staatsbosbeheer and IBED in the Oostvaardersplassen
concerning the effects of limiting grazing on nutrient and molecular cycles. Knowledge gained during both of these studies should be exchanged in order to broaden the understanding of large herbivores on the landscape in rewilding areas. Finally, I plan to establish
correspondence with dr. A. Torres to inquire about her rewilding scoring system, and with Amalia Llano to learn more about the interplay between rewilding success and geodiversity.
KNOWLEDGE UTILISATION
Beneficiaries
The findings from this research concerning the possible link between rewilding potential and geodiversity may prove helpful to organisations like Rewilding Europe, the Yellowstone to Yukon
Conservation Initiative and Rewilding Britain. These organisations can use this as a tool to determine which areas are the most promising candidates for a
rewilding approach. Education
In order to properly perform the required analyses on Yellowstone National Park, I will educate myself on the inner workings of the NDVI as well as Google Earth Engine.
Outreach method
The complete thesis will be uploaded to the University of Amsterdam’s online thesis library. This is an online
repository containing all BSc. and MSc. theses, which is freely accessible to all members of the public. As of now, there is no specific plan to advertise or spread awareness.
Data management
The data collected during the study will be assigned relevant and identifiable metadata. A copy will be held on my local computer, which is backed up on a daily basis to a Google Drive account. Interested parties will be provided with a link to access and read the online version of the data.
Data distribution
After the study has finished, the complete thesis is uploaded to the freely accessible thesis library of the University of Amsterdam. The resultant datasets are uploaded to the Open Science Framework, where they are easily accessible to interested parties.
EQUIPMENT
In order to properly perform the proposed research plan, I need several key pieces of equipment – mostly digital. The following is a list of required equipment:
Computer: since all analyses are performed
digitally and utilize digital data, a computer that at least conforms to the minimal specifications is required. My current laptop will fulfill this role, and past performance
indicate that there should be no serious issues with executing the necessary analyses.
ArcGIS license: in order to perform the
various geospatial analyses, an ArcGIS license is necessary. The current academic license I own is functional until the end of 2021, which is more than enough to cover the research period.
17
Matlab license: to perform the weighting and
validation of the WGDI, as well as the random selection required in several steps of the research, I need a Matlab license. My current license is active until July of 2021, which is enough time to cover the research period. In case of unforeseen setbacks, the optimization script has also been tested in the free software GNU Octave, and was found to act functionally identical.
Google Earth Engine access: to calculate
the NDVI of Yellowstone National Park and Lausitz, access to the services of Google Earth Engine is required. This is free for research purposes, but does require an admittance procedure. In case the research proposal is approved, I will submit my details to this process as soon as possible.
BUDGET
Since the research plan specifically aims to use open-access datasets and all the necessary equipment is already owned, the only expected costs are the salaries of the supporting staff members. Based on an hourly wage of €80,- per hour, the grand total amounts to €5960,- and is visualized in table 1. The different phases of the research likely contain different levels of involvement of the staff, leading to higher or lower costs per phase. During the preparatory phase, I expect to need an average of 2 hours per week of assistance from the examiner, and 30 minutes per week from the assessor. However, the NDVI and rate of change
calculations during this phase add another 10 hours to this. In the research phase, I expect to need an average of 2.5 hours per week from the examiner and 10 hours in total from the assessor for questions related to the spatial analyses. In the weighting and validation phase, I expect to need 2 hours from the examiner and 1 hour from the assessor. Finally, during the writing phase, I expect to need an average of 1 hour per week from the examiner and 30 minutes per week from the assessor, with the addition of 6 hours in total per individual for the assessment of the concept version of the thesis.
Table 1: an overview of the expected costs per phase of the proposed research trajectory.
Time prognosis Preparatory phase Research phase Weighting and validation phase Writing phase Total working hours 280 160 40 320
Examiner hours 14 10 2 14
Assessor hours 13.5 10 1 10
Associated salary costs (€)
Examiner 1120 800 160 1120
Assessor 1080 800 80 800
Total costs per phase (€) 2200 1600 240 1920
18
SAFETY
The entire research trajectory is designed to be performed by one individual working from home: all data should be open-access and retrievable from online databases, and all required software is either available on my private computer or freely available through online services. The current research design does not require fieldwork of any kind, so the associated safety risks are avoided. There might be a short fieldwork stint during the writing process, but this will be conducted on
foot in the Oostvaardersplassen in order to gather some image material to illustrate and enhance the thesis and will be performed by a maximum of three people. Due to the ongoing COVID-19 pandemic, consultancy with examiner, assessor, experts and researchers will primarily be conducted through video conference calls. In case physical encounters are necessary, all government-mandated precautions will be taken to ensure the good health of all involved.
19
REFERENCES
Benito‐Calvo, A., Pérez‐González, A., Magri, O., & Meza, P. (2009). Assessing regional geodiversity: the Iberian Peninsula. Earth surface processes and landforms, 34(10), 1433-1445.
Berger, J. (2009). The better to eat you with: fear in the animal world. University of Chicago Press.
Cardinale, B. J., Duffy, J. E., Gonzalez, A., Hooper, D. U., Perrings, C., Venail, P., ... & Kinzig, A. P. (2012). Biodiversity loss and its impact on humanity. Nature, 486(7401), 59-67.
Ceballos, G., Ehrlich, P. R., Barnosky, A. D., García, A., Pringle, R. M., & Palmer, T. M. (2015). Accelerated modern human–induced species losses: Entering the sixth mass extinction. Science advances, 1(5), e1400253.
Dirzo, R., & Raven, P. H. (2003). Global state of biodiversity and loss. Annual review of Environment and
Resources, 28.
Dirzo, R., Young, H. S., Galetti, M., Ceballos, G., Isaac, N. J., & Collen, B. (2014). Defaunation in the Anthropocene. science, 345(6195), 401-406.
Fechter, D., & Storch, I. (2014). How many wolves (Canis lupus) fit into Germany? The role of assumptions in predictive rule-based habitat models for habitat generalists. PloS one, 9(7), e101798.
Forte, J. P., Brilha, J., Pereira, D. I., & Nolasco, M. (2018). Kernel density applied to the quantitative assessment of geodiversity. Geoheritage, 10(2), 205-217.
Garroutte, E. L., Hansen, A. J., & Lawrence, R. L. (2016). Using NDVI and EVI to map spatiotemporal variation in the biomass and quality of forage for migratory elk in the Greater Yellowstone Ecosystem. Remote Sensing, 8(5), 404. Gray, M. (2005, January). Geodiversity and Geoconservation: what, why, and how?. In The George Wright Forum (Vol. 22, No. 3, pp. 4-12). George Wright Society.
Higgs, E., Falk, D. A., Guerrini, A., Hall, M., Harris, J., Hobbs, R. J., ... & Throop, W. (2014). The changing role of history in restoration ecology. Frontiers in Ecology and the Environment, 12(9), 499-506.
Hughes, F. M., Adams, W. M., Butchart, S. H., Field, R. H., Peh, K. S. H., & Warrington, S. (2016). The challenges of integrating biodiversity and ecosystem services monitoring and evaluation at a landscape-scale wetland restoration project in the UK. Ecology and Society, 21(3).
Kopnina, H. N., Leadbeater, S. R., & Cryer, P. (2009). Learning to Rewild: Examining the Failed Case of the Dutch “New Wilderness” Oostvaardersplassen.
Laundré, J. W., Hernández, L., & Ripple, W. J. (2010). The landscape of fear: ecological implications of being afraid. The Open Ecology Journal, 3(1).
Laurance, W. F., Useche, D. C., Rendeiro, J., Kalka, M., Bradshaw, C. J., Sloan, S. P., ... & Arroyo-Rodriguez, V. (2012). Averting biodiversity collapse in tropical forest protected areas. Nature, 489(7415), 290-294.
Llano, A. (2020). Rewilding success and physical features of the landscape: a spatial analysis of three European case study areas (Master’s thesis). University of Amsterdam, Amsterdam.
20
Lorimer, J., & Driessen, C. (2014). Wild experiments at the Oostvaardersplassen: Rethinking environmentalism in the Anthropocene. Transactions of the Institute of British Geographers, 39(2), 169-181.
Oliver, T. H., Isaac, N. J., August, T. A., Woodcock, B. A., Roy, D. B., & Bullock, J. M. (2015). Declining resilience of ecosystem functions under biodiversity loss. Nature Communications, 6(1), 1-8.
Perino, A., Pereira, H. M., Navarro, L. M., Fernández, N., Bullock, J. M., Ceaușu, S., ... & Pe’er, G. (2019). Rewilding complex ecosystems. Science, 364(6438).
Pettorelli, N., Barlow, J., Stephens, P. A., Durant, S. M., Connor, B., Schulte to Bühne, H., ... & du Toit, J. T. (2018). Making rewilding fit for policy. Journal of Applied Ecology, 55(3), 1114-1125.
Potter, C. (2019). Changes in vegetation cover of Yellowstone National Park estimated from MODIS greenness trends, 2000 to 2018. Remote Sensing in Earth Systems Sciences, 2(2-3), 147-160.
Ripple, W. J., & Beschta, R. L. (2007). Restoring Yellowstone’s aspen with wolves. Biological Conservation, 138(3-4), 514-519.
Schepers, F., & Jepson, P. (2016). Rewilding in a European context. International Journal of Wilderness, 22(2), 25-30. Soulé, M., & Noss, R. (1998). Rewilding and biodiversity: complementary goals for continental conservation. Wild
Earth, 8, 18-28.
Svenning, J. C., Pedersen, P. B., Donlan, C. J., Ejrnæs, R., Faurby, S., Galetti, M., ... & Vera, F. W. (2016). Science for a wilder Anthropocene: Synthesis and future directions for trophic rewilding research. Proceedings of the National
Academy of Sciences, 113(4), 898-906.
Thomas, M. (2012). A geomorphological approach to geodiversity-its applications to geoconservation and geotourism. Quaestiones geographicae, 31(1), 81-89.
Torres, A., Fernández, N., Zu Ermgassen, S., Helmer, W., Revilla, E., Saavedra, D., ... & Schepers, F. (2018). Measuring rewilding progress. Philosophical Transactions of the Royal Society B: Biological Sciences, 373(1761), 20170433. Vera, F. W. (2009). Large-scale nature development--The Oostvaardersplassen. British Wildlife, 20(5), 28. Yengoh, G. T., Dent, D., Olsson, L., Tengberg, A. E., & Tucker, C. J. (2015). Limits to the Use of NDVI in Land
Degradation Assessment. In Use of the Normalized Difference Vegetation Index (NDVI) to Assess Land Degradation at Multiple Scales (pp. 27-30). Springer, Cham.
