Prof. Dr. Carsten F. Dormann

Telefon: +49 761 203-3749
Telefax: +49 761 203-3751
eMail: carsten.dormann@biom.uni-freiburg.de 




Frau Eva Meier

Telefon: +49 761 203-3749 
Telefax: +49 761 203-3751 
eMail: eva.meier@biom.uni-freiburg.de 




Biometrie und Umweltsystemanalyse

Albert-Ludwigs-Universität Freiburg

Tennenbacher Straße 4 
79106 Freiburg i. Br.


You are here: Home Lehre / Teaching Vorherige Abschlussarbeiten / Past Theses Vorherige Abschlussarbeiten / Past Theses
Document Actions

Vorherige Abschlussarbeiten / Past Theses

Department theses defended later than 2012.

Bachelor Thesis

Disentangling the effects of light, silvicultural management and soil on forest plant diversity


Lukas Müller: Disentangling the effects of light, silvicultural management and soil on forest plant diversity
(2018): resolveuid/17311c76ba2c3e3acaf9406278284e32
BetreuerProf. Dr. Carsten Dormann 

Biodiversity is often understood to be the main driver of forest ecosystem processes and services. Much research from different perspectives have been done on this matter. The underlying thesis embeds these perspective into a model, while focusing on the yet not to be analyzed relationship between light and species diversity in the herb layer over species; light demand. My studies are based on, and enabled by an extensive research project on biodiversity called The Biodiversity Exploratories (Fischer et al., 2010), from which I considered 55 monoculture forest plots in three different regions in Germany of 1 hectare each. Relevant data from them was extracted and put into context, using structural equation modeling (SEM) as main approach for analyzing and bayesian modeling for validation. I consecutively tested multicollinearity, over-fitting after modi€cation of indices, while observing resulting explained variance and goodness of €t to reduce an initially complex model in order to fi€nd the simplest, yet still theory and data fiting one. the fi€nal model contains seven forest ecosystem properties: Soil type (ST), measured by sand content, main tree species as a measure for forest structure (FS), basal area share of forest canopy, indicating tree layer (TL), mean weighted Ellenberg light values representing properties of the herb layer (HL) as well as an index for forest management intensity ForMI (FM) (Kahl & Bauhus, 2014). Interrelating effects between these entities and on the environmental condition respectively main model variable biodiversity (BD) in the herb layer, assessed by effective number of species, especially from Light (LI), could be shown. My model is validated by its €t (P(X2): 0.065, RMSEA: 0.12, SRMSR: 0.17, CFI:0.96) and explained variance of dependent model variables (R2 SEM: HL: 0.84, TL: 0.82, LI: 0.5, FS: 0.82, BD: 0.63). ‘e la‹er is con€rmed by the same model with a bayesian a‹empt (R2 Bay.M.: HL: 0.7, TL: 0.81, LI: 0.45, FS: 0.42, BD: 0.57). It was performed to compensate for possible threads to SEM when using small sample sizes. According to my model, sand content in soil decreases biodiversity by 0.6 less plant species in the herb layer for every tenth of weight proportion. My results support the controversial assumption that forest management enhances plant diversity (4.8 more species for every ForMI index point). Furthermore, I discovered that for each percentage of photosynthetically active radiation of the key €gure of light passing through the forest canopy, 0.78 more plant species grow in the herb layer. As expected, the link is mediated by species; light demand. This coherence, among other known effects, resulting from my model, has not been previously demonstrated. How these fi€ndings will be used in future studies or practice goes beyond the scope of this thesis.


Improving the Global Forest Biodiversity-Productivity Relationship


Helge David Schneider: Improving the Global Forest Biodiversity-Productivity Relationship
BetreuerProf. Dr. Carsten Dormann 

The quantitative examination of the global forest biodiversity-productivity relationship (BPR) is essential to international forest management and the conservation of biodiversity. Most recently, a study by Liang et al. (2016) showed a positive and consistent relationship between tree diversity and ecosystem productivity. They used an extensive data set of 777,126 permanent plots, covering most of the terrestrial biomes. However, bearing in mind the great environmental variability across the world’s forests, the width of the confidence interval (CI) in their study seems strikingly narrow. I suggest that this is due to the calculation of the CI by the standard error of the mean, sampling errors, the rescaling of species richness and several other aspects. By applying various modifications individually to the analysis of Liang et al. (2016), I addressed these issues step by step. Changes included the recalculation of the CI, stratified sampling and a different rescaling of the species richness along with other secondary adjustments. All modifications were merged into a new model. The results suggest, that the error margin of the BPR is more than 20 times higher than stated by Liang et al. (2016) and the effect of biodiversity on productivity depends considerably on the local environmental conditions. Moreover, the findings indicate that a loss in biodiversity leads to a greater decline in productivity than stated in Liang et al. (2016), especially in forests with low biodiversity. Nonetheless, the new model corresponds with the main findings of Liang et al. (2016) and is supporting evidence that biodiversity in general has a positive effect on forest productivity.


Biome-specific analysis of the relationship between biodiversity and productivity in forests


Jonas Gorges: Biome-specific analysis of the relationship between biodiversity and productivity in forests
BetreuerProf. Dr. Carsten Dormann 

Worldwide economic and political descissions are made with the help of scientific perceptions. In times of raising populism and a general decline in confidence in all public institutions, especially studies that deal with highly emotional topics like biodiversity, has to be as precise as possible. Therefore, it is essential to continuous reexamine, extend and improve influential studies, models and allegations. This thesis aims for an further development of the study of Liang et al. (2016) on the relationship between tree species richness and productivity in forests worldwide. The global perspective is evaluated and compared with a biome-specific approach and the trade-off between a single large-scale model with higher uncertainty and multiple regional models with lower clarity is discussed. The implementation of changes includes both technical and conceptional approaches. The technical development of the model basically consists out of an enormous reduction of the calculation time, conceptional progression is implemented by an adjustment in the sampling method and a relativization of the response variable.



Entwicklung eines softwarebasierten Lösungsansatzes zur Förderung von Transparenz und Optimierung von Bestellabläufen in Food Co-ops


Simon Pöschl: Entwicklung eines softwarebasierten Lösungsansatzes zur Förderung von Transparenz und Optimierung von Bestellabläufen in Food Co-ops
BetreuerProf. Dr. Carsten Dormann 

Im Gegensatz zu konventionellen Lebensmitteldistributionsstrukturen legen kleinere, dezentrale Versorgungsorgane wie Food Hubs oft ihren Fokus auf die Nachvollziehbarkeit der Lebensmittelherkunft und den Bezug der Produkte von regionalen Kleinbauern1. Als eine Unterkategorie von Food Hubs reduzieren Food Co-ops den Zwischen- und Einzelhandel über den Weg des Zusammenschlusses von Verbrauchern. Daraus resultierend sehen sich Food Co-ops konfrontiert mit Herausforderungen, von denen das Finden geeigneter technischer Unterstützung die Ursache dieser Arbeit darstellt. Um diese Hürden abzubauen, wurde 2015 mit der Programmierung einer Android App, FoodCoApp, begonnen, die vorrangig als Warenwirtschaftssystem fungiert und seither in der Food Co-op Kornkammer e.V. in Freiburg eingesetzt wird. Die vorliegende Arbeit verfolgt das Ziel, Informationen überWarenbestände aus dieser App weiter zu verarbeiten und durch eine benutzerfreundliche Visualisierung der Daten die Transparenz und Bestellabl¨aufe innerhalb des Kornkammer e.V. zu verbessern. Die FoodCoApp erstellt im Rahmen eines regelmäßigen Backups eine Tabelle, die die tägliche Warenbestands¨anderung eines jeden Produkts enthält. Mittels der Programmiersprache R wurden Funktionen geschaffen, mit deren Hilfe sich spezifische Informationen aus der Tabelle in Form von Diagrammen und Plots visualisieren lassen. Zur besseren Benutzerfreundlichkeit und Bedienbarkeit wurden die Funktionen in eine shiny App gepackt - ein von Rstudio entwickeltes Werkzeug, um Grafiken interaktiv aufWebseiten anzuzeigen. Das Ergebnis der Arbeit ist ein R-Paket, foodstorage, das Funktionen enthält, mittels derer sich sowowhl der derzeitigeWarenbestand als auch der zu erwartende, zukünftigeWarenbestandsverlauf visualisieren lässt. Des weiteren erlaubt die shiny App, storage-app, das Abrufen der visualisierten Daten in einemWebbrowser. Sofern die letzten Aufgaben gemeistert sind, bevor die storage-app in der Praxis Anwendung finden kann, wird sie Mitglieder im Kornkammer e.V. ein detaillierteres Bild über den derzeitigen Warenbestand verschaffen und so zu einer Verbesserung der Transparenz innerhalb der Kooperative beitragen. Zudem lassen sich durch die Anwendung der App Bestellabläufe zu optimieren, indem durch Ermittlung von Zukunftsprognosen für den Lagerbestand frühzeitig neue Bestellzyklen eingeleitet werden können.Werden zudem in der FoodCoApp weitere Schnittstellen zu anderen open source Projekte geschaffen, so kann dieses Projekt auf großflächige Resonanz bei Food Hubs stoßen und so seinen Beitrag für mehr Resilienz regionaler Versorgungsstrukturen leisten.

Evaluating Measures of Phenological Overlap


Florian Berger: Evaluating Measures of Phenological Overlap
BetreuerProf. Dr. Carsten Dormann 

Contrary to the former notion that barnacle geese (Branta leucopsis) are obligate Arctic breeders, they have rapidly expanded their breeding range towards temperate regions in the last 50 years, resulting in the emergence of non-migrating breeding populations. This development raises questions about the mechanisms of adaptation to different environmental conditions, about changes in trade-offs when following new alternative versus traditional lifestyles, and about the resulting fitness consequences. Growth and body size are fundamental life history traits, and conditions young growing geese experience during the first months of their life may be decisive for their later performance later in life. An accurate knowledge of the variation of growth and the influencing factors is of great interest in evolutionary ecology and enables to confirm theories on adaptations to different and changing environmental conditions. Body size measurements are therefore an appropriate tool for studying variation between and within populations. In this investigation, nonlinear mixed-effects modelling of morphometric measurements from juvenile barnacle geese from an Arctic and a temperate breeding population has been applied in order to quantify the influence of age (from hatch close to fledging), breeding population, sex and hatch date relative to the colony-specific mean on the growth of body mass and three size measurements (head, tarsus and wing length). The main effects of breeding population, sex, and relative hatch date were all significant, meaning that geese grew faster in the Arctic, that males reached a larger asymptotic size than females, and that growth rates tended to be the higher the earlier the hatching date. The interaction effect of relative hatch date and population was likewise significant.


Exploring growth trajectories of barnacle geese from Arctic- and temperate-breeding populations using nonlinear mixed-effects modelling


Boas Steffani: Exploring growth trajectories of barnacle geese from Arctic- and temperate-breeding populations using nonlinear mixed-effects modelling
Betreuer: Prof. Dr. Florian Hartig


Contrary to the former notion that barnacle geese (Branta leucopsis) are obligate Arctic breeders, they have rapidly expanded their breeding range towards temperate regions in the last 50 years, resulting in the emergence of non-migrating breeding populations. This development raises questions about the mechanisms of adaptation to different environmental conditions, about changes in trade-offs when following new alternative versus traditional lifestyles, and about the resulting fitness consequences. Growth and body size are fundamental life history traits, and conditions young growing geese experience during the first months of their life may be decisive for their later performance later in life. An accurate knowledge of the variation of growth and the influencing factors is of great interest in evolutionary ecology and enables to confirm theories on adaptations to different and changing environmental conditions. Body size measurements are therefore an appropriate tool for studying variation between and within populations. In this investigation, nonlinear mixed-effects modelling of morphometric measurements from juvenile barnacle geese from an Arctic and a temperate breeding population has been applied in order to quantify the influence of age (from hatch close to fledging), breeding population, sex and hatch date relative to the colony-specific mean on the growth of body mass and three size measurements (head, tarsus and wing length). The main effects of breeding population, sex, and relative hatch date were all significant, meaning that geese grew faster in the Arctic, that males reached a larger asymptotic size than females, and that growth rates tended to be the higher the earlier the hatching date. The interaction effect of relative hatch date and population was likewise significant.


Soil moisture fluctuations and plant diversity: testing the hydrological niche trade-off hypothesis


Victoria Kolodziej: Soil moisture fluctuations and plant diversity: testing the hydrological niche trade-off hypothesis
BetreuerProf. Dr. Carsten Dormann 


This thesis analyzes the correlation between soil moisture fluctuations and the species richness of grassland plants in three German regions. A total of 120 study sites, located in the Schwäbische Alb, in the Hainich-Dün region and in the Schorfheide, were examined. The Biodiversity Exploratories project provided hourly soil moisture values and data about the occurrence of the individual plant species, the land use and the soil properties. Past studies determined that plant species are either well adapted to water logging or to dry periods, but not to both. So far it is unknown whether the hydrological niches are correlated to the plant species number. It was assumed that the niche space and thus the plant species richness increases with the soil moisture fluctuation. The extent of exceeding the water logging and drought thresholds can be quantified by using sum excellence values. The sum exceedance values for water logging (SEVa) and drought stress (SEVd) were determined for several years, starting 2010 until the end of 2013. First of all, the theory about the adaption of the plant species to the soil moisture conditions was investigated. Contrary to the expectations, the adaption to dry periods increased with the adaption to wetness. Furthermore, the correlation between the soil moisture fluctuation and the plant species richness was not confirmed. Neither the SEVd nor the interaction between both SEVs had a significant influence on the plant species number. Only the SEVs showed a significant correlation. A structural equation model was used to add the land use intensity, the humus content and the clay content of the mineral soil as predictors. The correlation between the SEVa and the plant species number was not significant within the structural equation model.


The correlation between niche breadth and range size: result of niche or neutral dispersal limitation?


Johanna Dodillet: The correlation between niche breadth and range size: result of niche or neutral dispersal limitation?
Betreuer: Prof. Dr. Florian Hartig


What determines species’ range sizes is a big issue of macroecology. There are different theories about it. One idea is, that the range size is driven by the ecological niche. Another idea is, that the range size is a result of neutral dispersal processes. Also it isn’t clear what exactly the niche is. Here, I examine the relationship between climatic niche breadth and range size for different numbers of niche axis by using data from the International Union for Conservation of Nature (IUCN) and WorldClim. Further I compare the relationship with two null models, to test what is the process that influences the relationship. In the first model, the range sizes are only dispersal driven and in the second model the range sizes are only niche breadth driven. My results are, that there is, for terrestrial mammals, a correlation between niche breadth and range size, which decreases with a growing number of niche axis. Secondly, this correlation is different from the relationship of both null models. In conclusion, this study supports the long assumed correlation of niche breadth and range size and shows that neither the niche breadth nor the dispersal rate alone can explain this correlation, but the reality is a combination of both. These results contribute to the understanding of why some species are rare and others widespread.


Douglas-fir growth under future climate: applications of a century of provenance trial data


Raphael Haber: Douglas-fir growth under future climate: applications of a century of provenance trial data
BetreuerDavid R. Roberts



Seed transfer and the suitability of tree provenances to the target climate are im-portant factors for reforestation and forest management issues. These aspects are expected to gain significance considering the imminent era of anthropogenic climate change. One of the most common practices in this field of research involves re-sponse functions that calculate growth as a function of climate variables. In this bachelor thesis, I applied response functions to European forests by using data from 112 common garden experiments to investigate the future growth of Douglas-Fir and its multiple provenances under different climate change scenarios. The main findings of this thesis are that Douglas-Fir is likely to stay a promising tree species for forestry in Europe, particularly at high altitudes and latitudes. According to my models, the optimal selection of planting provenances is not going to change significantly, even though we can perceive a trend towards seed material from dryer origins. Further-more, in terms of methodology I conclude that GLMs are a more suitable tool for modelling future tree height than predictions made with Random Forests models as a comparison.


Th‡e eff‚ect of competitive release on individual-level specialization


Kristina Dabrock: Th‡e eff‚ect of competitive release on individual-level specialization
Betreuerin:Gita Benadi


ThŒe concept of ecological niches is a fundamental principle of population ecology and evolutionary theory and has therefore been the subject of a variety of studies. One specifi€c subtopic is that of competitive release, the expansion of population niches through a reduction of interspeci€c competition. Against the background of diff‚erent studies coming to divergent conclusions as to the e‚ffect of competitive release on individual-level specialization, this paper aims at revealing the underlying principles and exposing the conditions under which individual niches are altered. Œe experiments are conducted by use of an individual-based model, which is implemented in the programming language NetLogo. ŒThey show a broadening of population niches combined with increased individual-level specialization and higher levels of inter-individual variation for certain parameter combinations under €xed population sizes, while individual-level specialization remains unchanged under variable population sizes. Œe specialization of individuals in the €first case is due to increased resource densities. ŒThey result in shorter search times and therefore allow individuals to invest extra time in the search for food items providing them with high amounts of energy. In the laŠer case, interspecifi€c competition is replaced by intraspecifi€c competition, resulting in constant grazing pressure and resource densities and therefore unchanged niches. Œus this study generally supports the theory of niche expansion, which states that population niches become broader when competitors are removed as well as niche variation hypothesis, which postulates that populations with a wider total niche exhibit higher inter-individual variability rather than wider individual niches. It also demonstrates that the decoupling of individual and population niche can be even stronger than suggested in former studies.



Nestedness and null models in ecology


Lara Kaiser: Nestedness and null models in ecology
(2015): resolveuid/08fe8b67ebb03eea029d37a2105f3622

Detecting patterns of interaction in networks is a primary goal of ecological research and a nestedness pattern has been detected frequently over the last years. Nestedness is a interaction pattern of a network that shows high occupancy of links in the upper left triangle of the ordered interaction matrix. As this shape may be the result of environmental gradients, it can be used to detect which gradients are those shaping the community and is therefore a very interesting ecological pattern. While there are many ways of interpreting a nested pattern, there are just as many ways of calculating it. More and more null models and metrics are developed, while their performance was not sufficiently tested for. While some metrics as well as some null models have been compared, there are still many unsatisfying evaluated possible combinations. It was, for example, not tested sufficiently for the influence of different matrix types. The possibility of method combinations leads to a problem in comparing the “nestedness” that was conducted by these different methods. The question arises whether different calculated nestedness even describes the same pattern. Differing in matrix ordering process, choice of metric and null model, ecologist have many options when it comes to evaluating data on whether or not it is nested. Therefore it is di￿cult to compare and infer from the result of nestedness analysis, if the processes in which this “nestedness” was created vary in the underlying methods. Existing literature on this topic was summarised to give a introduction on common metrics and null models and the results of previous analysis on their performance. Seven null models and ten metrics were chosen to test how they perform together. Intrinsic metric scaling was compared before applying the null models and was found to be inconsistent in their return. The same was done after applying the null models, which lead to even less consistency. Metric and null model performance on moduled matrices was analysed and were all found to be unsuited for such an analysis. One very constraint model was analysed on how different metrics influence it. I found that the metrics are not consistent in their definition of a nested pattern. Especially the T metric returns results very different from all other metrics.


Scale-dependent resource selection by Alpine ibex (Capra ibex) in summer


Milena Zurmühl: Scale-dependent resource selection by Alpine ibex (Capra ibex) in summer
Betreuer: Dr. Simone Ciuti

As not every habitat unit provides the adequate mixture of environmental conditions, animals often have to face a certain trade-off between several factors when selecting their habitats. Thereby, the main drivers, which need to be weighed against each other, are mostly food, safety and reduced exposure to unfavourable climate conditions. To find out how male ibex select their habitat and which seasonal decisions they make when confronted with the trade-off between the named factors, their resource selection behaviour has been observed in Gran Paradiso National Park (Western Italian Alps) during the growing season (May-October) of the years 2010 and 2011. The analysis was carried out on two different scales by means of resource selection functions (RSFs). The large scale was defined by the home range size of the whole population, which was determined by the calculation of a minimum convex polygon (MCP 100%). The smaller scale was described by the areas which male ibex could reach within a month. To consider that habitat use patterns vary also in time, both daily and seasonal parameters were included in the analysis. And in fact seasonal dependent selection patterns could be clearly detected for both scales. Among the different spatial scales however, only slight differences could be noticed. Over all, the trade-off between reducing the exposure to unfavourable climate conditions, especially to heat, and maximizing the energy input by foraging high quality food played the key role in the resource selection of male ibex. Most of the other environmental parameters seemed to only be selected in consequence to the seasonal dependent weighting of those factors. Additionally it was clearly shown that male ibexes rather react in expectations of the maximum air temperature values than in consideration of to the actual conditions, which indicates a certain predictive capability in terms of imminent temperature conditions. Regardless of the food quality and the highest temperature values male ibex preferred areas with high forage quantity. The reduction of predation risk however did not seem to have a huge impact on the resource selection of those animals.


Auswirkungen energetischer Gebäudesanierung auf ein Verteilnetz - Bewertung anhand eines Optimierungsmodells


Franz Irrenberg: Auswirkungen energetischer Gebäudesanierung auf ein Verteilnetz - Bewertung anhand eines Optimierungsmodells
(2015): resolveuid/58599f1dcabdc39b091bd1a059409e8a

Die vorliegende Arbeit wurde als Teil des Forschungsprojektes "StroWae – Analyse einer Marktrolle zur Optimierung der dezentralen Strom-Wärmenutzung" am Fraunhofer Institut für Solare Energiesysteme verfasst. Die Zielsetzung besteht darin die Auswirkungen unterschiedlicher Sanierungsstandards verschiedener Wohngebäude auf ein Wärmeverteilnetz zu analysieren. Dazu wird der Gebäudebestand des Verteilnetzes analysiert und eine Einteilung in repräsentative Referenzgebäude vorgenommen, sowie Sanierungsstandards nach den Mindestanforderungen zu den Wärmedurchgangskoeffizienten der Gebäudebauteile durch die gesetzliche Energieeinsparungsverordnung und einem Passivhaus-Dämmstandard des Förderprogramms der Kreditanstalt für Wiederaufbau gewählt. Auf Basis dieser Ausgangsdaten wurde mit einem statistischen Lastprofilgenerator des Fraunhofer ISE der Wärmebedarf der Referenzgebäude in allen Sanierungsstandards bestimmt, anhand dessen ausführlich die Energieeinsparung durch die Sanierungen untersucht wird. Mit dem Wärmebedarf als Input wird durch ein projektinternes Modell ein optimales Versorgungssystem minimaler Kosten in zwei Szenarien für alle Referenzgebäude in dem jeweiligen Sanierungszustand berechnet. Das erste Szenario basiert auf einer Fernwärme-Ausgangsversorgung der Referenzgebäude und liefert als Ergebnis eine von den Sanierungen unabhängige, fast ausschließlich mit Solarthermie betriebene Wärmeerzeugung, worauf hin ein weiteres Szenario mit derselben Ausgangsversorgung, aber einer Erhöhung der solarthermischen Kostenparameter um die eines Wärmespeichers betrachtet wird. Die Erhöhung der Kostenparameter bewirkt keine Veränderung in der Konstellation der zur Deckung des Wärmebedarfs eingesetzten technischen Anlagen, hat aber einen starken Einfluss auf die Zusammensetzung der gesamten Systemkosten. Für beide Szenarien kann festgestellt werden, dass die Gesamtsystemkosten eine Reduktion gleicher prozentualer Höhe durch die Sanierungen erfahren und die Gesamtsystemkosten im Wesentlichen durch die in der Betriebsführung der technischen Anlagen anfallenden Kosten bestimmt werden.


Seasonal habituation to human activities in alpine marmots (Marmota marmota) in the Gran Paradiso National Park, Italy


Julia Greulich: Seasonal habituation to human activities in alpine marmots (Marmota marmota) in the Gran Paradiso National Park, Italy
Betreuer: Dr. Simone Ciuti

Extreme weather conditions and short vegetation periods minimize the time that alpine animals can spend on feeding and reproduction. Moreover high tourist numbers, and the associated human disturbance, in alpine regions can have a negative impact on the fitness level and reproduction rate of animals since fleeing has costs in addition to the benefits of reducing the risk of predation. Hence habituation is a good response by animals of not wasting energy and time that could have been spent on other vitally activities. We used alpine marmots (Marmota marmota) in the Gran Paradiso National Park, Italy, to understand seasonal habituation. Therefore we analysed flight initiation distances (FID, the distance at which an animal starts to flee from an approaching observer) in relation to the distance to different types of hiking trails in September 2014 and June 2015. We expected that FID would decrease the closer a marmot is being situated to a high frequented main track, because habituated animals show reduced flight distances. We created an analysis framework, using a generalized additive model, and found that: (1) FID is significantly lower the closer a marmot is being situated to a high frequented trail (2) there is no significant difference in FID of September 2014 and June 2015 (3) FID is significantly higher, if the den is between the marmot and the observer or on the side (4) FID is significantly lower at midday and (5) young individuals flee significantly later than adults. Marmots thus responded to some human disturbance by adjusting their flight behaviour. Lower FID close to high frequented trails lead to the assumption that those individuals are able to adapt to predictable human disturbance and therefore are habituated. Furthermore the level of habituation is comparable prior (September 2014) and after (June 2015) hibernation. These results should be taken into account when developing management plans for remote alpine areas. In addition our approach will be useful to show that the distance to the closest trail is, according to our results, the main driver for habituation.


A statistical approach to species-area relationships for range data


Karina Reckling: A statistical approach to species-area relationships for range data
(2015): resolveuid/6fa6a1f5ba683fd2aef9c21b300d9aa1

ThŒis thesis is analyzing the scale-dependence of the z-values of the species-area relationship (SAR). Œe increase in the number of species when adding of area is considered a general law in ecology. Within the discussion of SAR the terms species richness, species density and species diversity are used interchangeably, although their meanings as defi€ned by WhiŠaker vary. Richness refers to the number of species per sample, whereas density refers to the number of species per sampling area. Diversity, meanwhile is divided into inventory diversity and di‚fferentiation diversity. A common interpretation of the diff‚erentiation diversity is the beta diversity, which is considered to be the rate of rise of the SAR linear relationship in logarithmic space, the z-value. Œe scale-dependence of this parameter is debated intensely. Most studies, such as this one, use a log-log transformation of the power law as introduced by Arrhenius in 1921: log(S) = log(c) + z  log(A) Œis mathematical function was applied on the rodents data of the Terrestrial Species dataset of the International Union for Conservation of Nature (IUCN) for diff‚erent levels of nestedness. Grid sizes of 50 km by 50 km, 100 km by 100 km, 200 km by 200 km, 500 km by 500 km and 1000 km by 1000 km and the MCMCglmm analytical model of the R statistical so‰ware was used. Also subsets of 500, 1000, 5000, 10000, 15000, 18059 and 36118 random samples were analyzed. Œe results of this process showed little variation within the di‚erent levels of nesting, for which the z-values remain in the magnitude of 0.1. A diff‚erence of one magnitude was detected for z-values of analysis without any nesting; here they were at 0.01 magnitude. For the c-values liŠttle variation was found. Œis diff‚erence in magnitude shows the scale-dependence of z-values. Since biological consequences and the thus missing linearity are not studied here, further reading and research is encouraged.

Identifying Coexistence Mechanisms from time series data


Christian Buschbeck: Identifying Coexistence Mechanisms from time series data
(2015): resolveuid/b40239fcad8e910c5d3229191431cfbf
Betreuer: Dr. Florian Hartig
Source code

In ecology, coexistence mechanisms play a major part in explaining the maintenance of biodiversity. Many previous studies have examined the different mechanisms and their properties with the help of analytical models and simulations. While these studies concentrate on understanding one mechanism in detail, here the focus lays on the distinction between different mechanisms. The idea is to identify every one of them by looking at their times series data. Hence, time series were created, with a time discrete two species model. Then I suggest two different approaches to distinguish the mechanisms based on characteristic patterns that occur in the simulated population dynamics. For the differentiation, I introduce five assessment tools that quantify the observed patterns. The validation of this differentiation showed that it is possible, to significantly distinguish the mechanisms for fixed parameters. The combinations of coexistence mechanisms could be identified less significantly and not for all combinations. For alternating parameters, the significance was reduced as well. Even though, the assessment tools might still find use in indicating coexistence mechanisms and should be tested on field data. Further research needs to be done to examine the combinations of coexistence mechanisms.


Modelling Population Dynamics of North Dakota Mule Deer


Lennart Schmidt: Modelling Population Dynamics of North Dakota Mule Deer
(2014): resolveuid/77d2343f3b554f19db8a1bcf0feace36
Betreuerin: Dr. Gita Benadi

Mule deer are of value both for local communities as an economic factor and for research in population ecology due to their complex population dynamics. Knowledge about the factors that cause changes in population densities is not yet sufficient to make predictions about future trends and the pattern of research is too fragmented to allow for transferability of study results. In this study, I examined a study area in western North Dakota of about 7000 km2 regarding the direct effect of environmental factors on mule deer population densities. Two methods were applied. First, I performed regression analysis using ’Generalized Additive Models’. The results of this analysis were validated by fitting ’state-space’ models to the observed population densities. ’State-space’ models have recently been rediscovered in research and represent an approach that is both stochastic and deterministic. The primary result of both model types is that the main effect of increases in hunting on population densities is negative. Its magnitude varies with spatial location. The effect of woody vegetation was proven to be negative at rates of woody coverage below 21%. These results agree with previous research. Furthermore, I detected non-linear density dependence of the relative growth rates in the study area. This has been a major obstacle in previous studies, therefore representing a starting point for future examination of the interaction of density-dependence and other variables. These could include vital rates or environmental factors. The results are considered to be representative and transferable as they were obtained from a large area and could be backed with other research results. The ’state-space’ models proved to be powerful tools to supplement a stochastic analysis of population data. However, methodological knowledge needs to be present to exploit the whole potential of these models. In this study, common procedures of the evaluation of the results from ’state-space’ models could be improved. If extended in future projects, the ’state-space’ models that I set up in this study could serve as prediction tools for harvest management purposes – thus, positively influencing local communities that depend on the hunting sector. 

Exploring the ecological and evolutionary drivers of diversification of the alpine plant genus Phyteuma (Campanulaceae)


Laura-Marie Vecera: Exploring the ecological and evolutionary drivers of diversification of the alpine plant genus Phyteuma (Campanulaceae)
(2014): resolveuid/057b59feafd6f53676239567925161cf

On the evidence of current anthropogenic climate change it becomes increasingly important to understand how it could affect different species. If they are incapable of adapting by evolving innovative traits, they can only persist if they migrate to areas where the former environmental conditions still exist. Investigating the evolutionary history of certain genera can help us to gain a better understanding for these processes. The plant genus Phyteuma is a suitable study group for this purpose. It has been exposed to changing climatic conditions several times during the quaternary. Nevertheless, none of its species show traits that hint at an evolution of climatic niches. It rather seems like populations followed the alpine climate and new species evolved after geographic separation and consequently reproductive isolation. The aim of this study was to test these hypotheses to learn more about the evolutionary history of Phyteuma. After the reconstruction of the phylogeny, geographical and ecological attributes of sister species, who share a common ancestral history, were compared to those of non-sister species. Complete geographic separation and niche conservatism are considered as prerequisites of allopatric speciation. Thus, sister species should have a low range overlap and a high overlap of climatic niches compared to non-sister species. Climatic niches were estimated by climatic niche modelling. Furthermore, it was tested whether secondary contacts could have led to a higher range overlap between older sister species. As changing environmental conditions can be expected to divide populations randomly, the range asymmetry between species pairs should have no clear tendency. Projections of occurrence of Phyteuma for the Last Glacial Maximum and today were produced based on the climatic niche models to see whether the adaptation to colder climates could have led to a decrease of range size in the course of global warming. The results mainly confirm the hypotheses. Range asymmetry between species pairs was similar for sister species and non-sister species. The high overlap of climatic niches between sister species compared to non-sister species indicates that niche conservatism was prevalent in this genus. The predicted decrease of range size confirms this result. On the contrary, sister species had a higher range overlap than non-sister species. This pattern could emerge from secondary contacts as older sister species had a higher range overlap than younger ones. P. spicatum and P. gallicum seem to have evolved under sympatric speciation. P. pyrenaiucm and P. vagneri have clearly undergone a long distance dispersal event. In general, the results of this study indicate that the alpine plant genus Phyteuma predominantly diversified through allopatric speciation. Changes to warmer climates and the lack of capability for adaptation likely forced populations to track their preferred climate. Apparently, secondary contacts occurred during colder periods and could explain the co-existence of formerly allopatric species. To finally confirm all of these results, further research is necessary

Positive und negative Dichabhängigkeit als Mechanism zur Erhaltung der biologischen Vielfalt. 


Antje Lessmann: Positive und negative Dichabhängigkeit als Mechanism zur Erhaltung der biologischen Vielfalt.
(2014): resolveuid/acb759605968d0a3055d93bd804ea966
BetreuerinDr. Gita Benadi

Wie Arten mit änhlichen Anspruchen an ihre Ressourcen koexistieren können und die Diversität über einen langen Zeitraum erhalten bleibt, ist ein wichtiges Fosrchungsthema in der Ökologie. Durch theoretische Ansätze wird angenommen, dass die Vielfalt der Arten bestehen kann, wenn die Wachstumsrate einer Art mit ihrem zunehmenden Vorkommen gehemmt wird. Durch diese negative Dichteabhängigkeit kann keine Art übermäßig stark anwachsen, wodurch viele unterschiedliche Arten im selben Gebiet überleben können. Der gegentele Effekt ist die positive Dichteabhängigkeit. Vorteile für das Wachstum hat dabei diejenige Art, die schon häufig an einem Ort vorkommt. In der Natur ist dieses Phänomen beisplsweise bei Planzen mit spezialisierten Mykorrhiza zu beobachten. Die Pflanzen profitieren von der Nähe ihrer Artgenossen, da sie untereinander ein effizientes Hyphensystem aufbauen können. Positive Dichteabhäangigkeit wird bisher meist nur in den Zusammenhang mit dem Verlust von Arten gebracht und findet daher als Mechanismus der Arterhaltung noch wenig Beachtung in der Literatur. Dieser Betrachtungsweise setzen Molofsky und Bever (2002) ein räumliches Modell entgegen, welches zeigt, dassintraspezifische positive Dichteabhängigkeit die Erhaltung der Diversität über lange Zeiträume hinweg ermöglichen kann und durch die positive Interaktion bestimmte räumliche Muster entstehen. In dieser Arbeit wird das Modell zunächst nachgenildet und dann erweitert. Durch die zusätliche Implementation von negativer Dichteabhängigkeit, können die verchiedenen Interaktionen und ihr Zusammenwirken mit ihrer Auswirkung auf die Diversität und auf die Entstehung von räumlichen Mustern untersucht werden. Die Ergebnisse zeigen, dass positive Dichteabhängigkeit alleine, verglichen mit einem System ohne Dichteabhängigkeit des Wachstums, zur Arterhaltung über einen bestimmten Zeitraum führt. Im Modell erhält negative Dichteabhängigkeit die Arten vollständig bis zum Ende der Simulationszeit. In der Literatur wird diese Dichteabhängigkeit als Mechanismus der DIversitätserhaltung (Chesson, 2000) beschrieben. Im Vergleich dazu verzögert positive Dichteabhängigkeit das Aussterben der Arten. Wirken positive und negative Dichteabhängigkeit auf der selben räumlichen Skala, verhindert der eine Prozess den jeweils anderen positiven Effekt auf die Arterhaltung. Eine Zuordung der Interaktionsart anhand des räumlichen Musters ist daher kaum möglich


Monitoring eco-touristic activities: Turtle-watching boat trips in the National Marine Park of Zakynthos, Greece 


Friederike Stramm: Monitoring eco-touristic activities: Turtle-watching boat trips in the National Marine Park of Zakynthos, Greece 
(2014): resolveuid/8d68023f19722d91c6bfd12ba897d8fe
BetreuerinIsabel Avila

The National Marine Park of Zakynthos (NMPZ), Greece is the most important reproduction site for Loggerhead sea turtles (Caretta caretta) in the Mediterranean. In its coastal waters several operators are running turtle-watching boat trips. Management measures, such as a zonation scheme, a code of conduct and an endorsement scheme, are supposed to ensure a minimum impact of this activity on the species. So far, the effectiveness of these measures and the behavioural effects of turtle-watching have not been monitored. A long-term monitoring is essential to guarantee that the existing guidelines and zonation are - and will remain - suitable to protect this endangered species. The purpose of my thesis is to make an overall characterization of the turtle-watching activity in Zakynthos and to propose a methodology for measuring the ecological impact on sea turtles, i.e. their response to the presence of boats, as well as for measuring the effectiveness of the protection scheme, such as compliance with guidelines or the use of the distinct boating zones. Protection schemes for sea turtles in their marine habitat are in many cases, e.g. in seismic surveys, similar to those used for cetaceans. Therefore methods used to record animal behaviour characterize the business have been based on impact assessments and surveys carried out for whalewatching. Based on field work from May to October 2013 and on estimations derived from the available data, approximately 72,500 customers take part in turtle-watching trips on Zakynthos and generate an income of over 1.3 million EUR per year. The turtle-watching season in Zakynthos lasts from May to September with its peak in late July and early August which is the peak touristic season in Greece. The majority of the sea turtle nesting activity, according to the available monitoring data, takes place one month prior to this peak. To measure the behavioural effect of boating on sea turtles I propose for on-board observers to record for each turtle (1) the breathing frequency, (2) the predominant dive pattern, (3) displacement in direction of or away from the boat, (4) increase of speed displacement, (5) erratic behaviour or (6) change of travelling pattern. Also the observer should record the observation period, the type and activity of vessels as well as the number of boats and tourists taking part in turtle-watching. The data sheet for turtle encounters that I propose may be used by conservation volunteers or NMPZ employees and help to quantify the turtle-watching activity in the area and its effect on turtles. This may serve as a baseline to support future management plans of the eco-touristic activity and the conservation of sea turtles. 

Generalised Degrees of Freedom versus Cross-Validation as Measure of Model Complexity 


Severin Hauenstein: Generalised Degrees of Freedom versus Cross-Validation as Measure of Model Complexity (2014): resolveuid/6c1531b00dbe840e9f4b264ea3a5a0da

Model averaging is an alternative to standard model selection, which increasingly aŠracts attention in the €eld of ecology. Œe combination of model ensembles and machine learning methods makes a revision of standard model selection procedures necessary. In this analysis, I compare Akaike’s Information Criterion (AIC) based on generalised degrees of freedom (GDF) as measure of model complexity, with a 5-fold cross-validation. In oder to make this more applicable to ecological research, I primarily expand the concept of GDF to binary data and assess its performance for several con€gurations. In an overall comparison, I show that GDF-based AIC is approximately equivalent to a 5-fold cross-validated log-likelihood. Furthermore, I discuss cross-validation weights as an alternative to Akaike weights in model averaging and €nally, I propose a di‚erent measure of model complexity, derived from cross-validation. Œe analyses are fed by two simulated and two real life datasets, of which three are binomially distributed.

An Individual-based Model for the Distribution of Marine Mammals 


Martin Palt: An Individual-based Model for the Distribution of Marine Mammals (2013): resolveuid/40189fea81c81eb5edb7280bafa84b7e

Today the blue whale still raises questions about its habits of traveling the oceans. Different populations in different parts of the world share common patterns though even the combination of all those aspects does not fully reveal the mechanisms behind the whales’ long journeys. It remains uncertain if there are defined breeding grounds and there is evidence that migratory behaviour in fact does not apply to populations as a whole. The current state of science is summarized here and general concepts of small and large scale movements are illustrated. This includes all known present populations in their presumed migration tendencies as well as different modes of locomotion. Based on all this general rules for an agent or individual based model (ABM, IBM) are laid out and subsequently implemented in the NetLogo programming software. Results of initial runs of simulations are presented but they still lack in various areas. Nonetheless, they may evoke optimism about the model’s potential as a number of future developments is adapted. 

Random Walk Models to Explore Distribution Patterns of Marine Mammals


Matthias Hainmüller: Random Walk Models to Explore Distribution Patterns of Marine Mammals (2013): resolveuid/697bef22fe54774bba7e66da7c574088

Der Atlantische Nordkaper (Eubalaena glacialis) zählt zu den gefährdetsten Walarten der Welt. Die Wale stehen bereits seit mehreren Jahrzehnten unter Schutz. Trotzdem haben sich die Bestände kaum erholt. Problematisch ist, dass ein Drittel aller Todesfälle auf menschliches Handeln (Schiffsverkehr, Fischerei) zurückzuführen ist. Seine Migrationen entlang der Ostküste Nordamerikas konnten bislang nur unzureichend beschrieben werden, was den effektiven Schutz der Art erschwert. Um die Auswirkungen fehlender Beobachtungsdaten zu mindern, wird im Bereich des Artenschutzes oftmals auf Individuenbasierte Simulationsmodelle zurückgegriffen. Im Rahmen dieser Bachelorarbeit wurde ein Modell entwickelt, das die Migrationen der Nordkaper auf räumlicher und zeitlicher Ebene simuliert. Zur Modellierung des Walverhaltens wurde auf bekannte Korrelationen zwischen Walverteilung und verschiedenen Umweltparametern (Meeresoberflächentemperatur, Meerestiefe und Verteilung der Hauptbeute, Calanus finmarchicus) zurückgegriffen. Das Modell erstellte monatliche Verteilungskarten, die dann im Bezug auf beobachtete Werte ausgewertet wurden. Durch den Vergleich der Daten konnten Rückschlüsse auf die Qualität des Modells gezogen werden. Obwohl das Modell noch einige Schwachstellen aufweist, könnte es bereits als Grundlage für die Entwicklung weiterer Modelle verwendet werden.

Inferring competition from ocurrence patterns of amphibians 


Lukas Flinspach: Inferring competition from ocurrence patterns of amphibians (2013): resolveuid/54ab281e9921c88ffb4b37f48cec4bc7

In species communities, inter-specific competition leads to a segregated distribution of species. The segregation becomes apparent when a number of spatially distinct but otherwise similar ecosystems are compared to each other. In biogeographic studies, this is done in the analysis of so-called occurrence matrices. These binary matrices, consisting of sites and species, are commonly used. Abundance-based quantitative matrices, however, promise more detailed insights into community structure. Patterns in these matrices, such as segregation of species or nestedness, are quantified by specific metrics, and subsequently compared to a set of random or null matrices. The null matrices are created on the basis of the original matrix using a randomisation algorithm, the null model. This procedure tests whether or not the matrix pattern may have occurred by chance. Segregation of species in an occurrence matrix is generally tested for using the combination of a null model which maintains the row and column sums of the original matrix, and the Cscore metric, which counts the number of times two species replace each other on a pair of sites. Patterns of species segregation can be caused by several ecological processes, for example niche differentiation. As the detection of interspecific competition is the goal of this work, these alternative processes have to be controlled for. This case study attempts to discover inter-specific competition in a community of six anuran species colonising 179 ponds in a floodplain in central Germany. In eight species pairs, competition was expected due to similar breeding habitat preferences or phylogenetic relations. The geographic location of the sites and a list of environmental variables was investigated in order to detect possible causes of species segregation. Then, the community matrix was randomized using one binary and two abundance-based null models. The species appeared to use ephemeral ponds differently along the gradient of pond endurance time. Together with the geographical distribution of the species in the stud area, this variable gradient could potentially explain segregated patterns in three of the species pairs. The binary null model, however, revealed only one significant pattern of species segregation in a fourth species pair. As information on other potentially influential pond characteristics was missing in this study, however, this segregated pattern could not be fully identified as the result of inter-specific competition. The abundance-based null models were discovered to reveal patterns of intra-specific aggregation rather than inter-specific segregation. As the randomisation algorithm significantly increased matrix fill, the models were considered unsuitable for the detection of species segregation when combined with the binary metric C-score. Consequently, the use of a quantitative metric is recommended when dealing with quantitative data. In the study area, more exhaustive observations are required to identify environmental variables around which niche differentiation of the species might occur. 

The ideal seed dispersal distance under the impact of plan-plant interactions


Lukas Boersen:The ideal seed dispersal distance under the impact of plan-plant interactions (2012): resolveuid/3e84a38506e37cefa9b1f57e9184594e

Pflanzen entwickeln, angepasst an Populationsstruktur und andere Umwelteinfl¨usse, eine ideale Samenausbreitungsstrategie. Die Faktoren, die die Art der Strategie bestimmen sind vielf¨altig. Unter anderem spielen die Populationsstruktur, die Verteilung der Individuen im Raum und Pflanzeninteraktionen eine bedeutende Rolle. Neben der Konkurrenz um Platz, Licht oder Nährstoffe gibt es auch Interaktionen mit positiven Effekten. Diese Benachteiligungen und Begünstigungen hängen von verschiedenen Faktoren ab. Die Interaktionen spielen sich auf unterschiedlichen räumlichen Skalen ab, die sich in unterschiedliche großen Referenzbereichen widerspiegeln, deren Ausmaß den Einfluss des jeweiligen Effekts bestimmen. Diese haben wiederum bedeutenden Einfluss auf die Reproduktionsrate, die räumliche Ausbreitung der Population und somit eben auch auf die mittlere Samenausbreitungsdistanz. Ein grundlegendes Verst¨andnis der Prozesse, die die Entwicklung verschiedener Samenausbreitungsstrategien bewirken ist von großer Bedeutung für den Naturschutz. Pflanzen müssen ihre Samen durch die anthropogene Zerstückelung der Landschaft und daraus resultierender Habitatfragmentierung über immer größere Abst¨ande verbreiten. Für den Artenschutz ist es deshalb ungemein wichtig, die Prozesse der Samenausbreitung zu verstehen, um wirkungsvolle Naturschutzmaßnahmen zum Erhalt der Biodiversität zu entwickeln. Die vollständige Aufklärung der Verstrickung dieser Prozesse ist eine große Herausforderung f¨ur Evolutions¨okologen. Um die verschiedenen Auswirkungen von Pflanzen Interkationen zu quantifizieren wurde ein individuenbasiertes Modell entwickelt. Nachdem eine Parameterkombination ermittelt wurde, die den erwarteten Trade-off zwischen verschiedenen Samenausbreitungsdistanzen darstellt, wurden verschiedene Größenverhältnisse der Referenzbereiche, welche das Ausmaß des Einflusses von Konkurrenz und Beg¨unstigung bestimmen, getestet und deren Auswirkungen auf die Populationsgröße, die räumliche Struktur der Populationen und die mittlere Samenausbreitungsdistanz untersucht. Anhand der Ergebnisse konnten grundlegende Aussagen ¨uber die Auswirkungen der beiden Interaktionstypen getroffen werden, insbesondere wann die positiven oder negativen Effekte überwiegen können und einander kompensieren. Mit dem entwickelten Modell können weitere Experimente durchgef¨uhrt werden, deren Ergebnisse in empirischen Experimenten in der Natur getestet werden müssen, um deren Allgemeingültigkeit zu best¨atigen. Das Modell kann außerdem als Werkzeug dienen, um Experimente durchzuf¨uhren, welche mit den heutigen Untersuchungsmethoden noch nicht in der Realität möglich sind. 


Master Thesis


Investigating the effect of visitor frequency on ungulate distance to hiking trails in the Swiss National Park 

Jakob Viße: Investigating the effect of visitor frequency on ungulate distance to hiking trails in the Swiss National Park (2018)
Betreuer: Prof. Dr. Carsten Dormann

Recent years have seen a steady increase in outdoor recreation, especially to the most pristine areas, like national parks. This raises questions about the potential impact of recreational activities on wildlife inhabiting the park. Hence, the main challenge for national park management is balancing tourism and nature conservation. To assess the impact of visitors on wildlife, we investigated how daily visitor numbers affect distances of Alpine chamois Rupicapra rupicapra and red deer Cervus elaphus to hiking trails, in an alpine area that has been strictly protected for more than 100 years, including a hunting ban, and without large predators resident. We expected some degree of habituation and predicted that (1) animals stay closer to trails, when they were trapped compared to immobilized, (2) animals stay closer to trails later in the season and when they are older, and (3) animals stay further away from the trails, when they are in open habitat. We performed a geographical and statistical analysis, separately modelling GPS locations from 36 chamois and 20 red deer as a function of visitor numbers and several environmental and individual specific factors over a period of 9 years within an additive model framework. In the final models main effects of visitor numbers were zero and elevation was the most important predictor variable in both models, affecting the distance from trails the strongest. Moreover, modelled effects for both species differed noticeably. Predicted distances from trails had large margins of uncertainty. However, some trends and distinct patterns crystallised: chamois, generally, stay further away from trails than red deer; contrarily, there was virtually no difference in distance for both capture method and cover. We discuss our results with regard to common issues of spatial data analysis, and the dificulty of finding general patterns from studies on wildlife responses to human disturbance. Finally, we give some recommendations, how national parks can incorporate findings like ours in their management schemes, and emphasize the need for further research. 


Direct and indirect environmental control of abundance and diversity assessed by Bayesian structural equation models 

Florian Keppeler: Direct and indirect environmental control of abundance and diversity assessed by Bayesian structural equation models (2018)
Betreuer: Prof. Dr. Carsten Dormann

Applications of Bayesian structural equation modeling in ecological research are sparse, regardless of the potential of this statistical framework to asses direct and indirect effects of environmental control on abundance, diversity and community composition. Since these factors are closely linked to ecosystem key processes and properties, such as productivity, energy flow and geochemical cycling, Bayesian structural equation modeling might provide a much needed methodological approach to monitor and maintain ecosystem functionality. This scientific work presents an exploratory application of Bayesian structural equation modeling on large-scale data sampled in the course of the biodiversity exploratories project (Fischer et al.; 2010) consisting of 176 variables of various data types, with high ratios of missing observations within a sample size of 150 observed plots. Latent data generation is used to impute missing observations and to create latent variables within factor analyses, which are conducted to reduce data dimensionality within hierarchical levels. Interrelations between these latent variables are assessed by exploratory path analysis. The displayed results testify the suitability of Bayesian structural equation modeling to account for spatial effects, by the possibility to identify and asses main drivers, such as geological and climatic factors. Further, the presentation of results in the form Path diagrams are shown to be easily interpretable and the Bayesian framework is a convenient mean to achieve intuitive uncertainty estimations. The flexibility of Bayesian analysis software JAGS enabled the evaluation of heterogeneous data and is expected to integrate time series and more complex model designs. Current limitations of Bayesian structural equation modeling, manifested in the lack of a suited goodness of fit measure, relatively high computational effort and model bias due to regression dilution, are pointed out and possible solutions are discussed. 


Behavioral Responses of GPS-collared European Red Deer (Cervus elaphus) to Off-trail Recreation and Hunting 


Zeno A. Bader: Behavioral Responses of GPS-collared European Red Deer (Cervus elaphus) to Off-trail Recreation and Hunting (2017)
Betreuer: Dr. Simone Ciuti

Outdoor recreational activities, both on-trail and off-trail, have been growing in popularity over the last decades. These trends cause a spatial and temporal expansion of recreational activity, which results in a higher amount of potential disturbances for wildlife. In this study, GPS telemetry and activity sensor data from a total of 45 red deer from three different areas (Southern Black Forest, Bavarian Forest National Park and Eastern Switzerland) was analyzed. It was assessed if human interactions cause changes in spatio-temporal behavior of red deer by: proportion of time being active, Euclidean distance and the distance to human infrastructure. For this purpose, off-trail disturbance experiments were conducted during summer (simulating mushroom- or berry-pickers). Additionally to the disturbance experiments, drive hunts and an exemplary winter interaction were analyzed in the Southern Black Forest. The results revealed an increase of activity for all types of human red deer interactions (HRI), except for drive hunts. All types of HRI evoked an increase of the Euclidean distance on the day of the HRI, especially on the day of the HRI in winter. Distance to human infrastructure was not influenced by the encounters with humans, but revealed daily, as well as seasonal differences. The activity data, as well as the Euclidean distance seem to be proper measures to assess short term responses. But it was only possible to examine long-term effects on red deer, caused by human red deer interactions, by using activity data. In conclusion, it seems that the best strategy to minimize human disturbance for red deer, is a predictable human presence, hence that recreational activities only take place on marked trails


Identifying Ecotypes from Distribution Data 

Marieke Wesselkamp
: Identifying Ecotypes from Distribution Data (2017)
Betreuer: Prof. Dr. Carsten F. Dormann

In applied ecology, species-habitat associations are known to vary spatially for widely spread species. An issue that recently received more attention when simulating a species distribution from environmental data is the non-stationarity of the modeled relationships. Often, ecological relationships are modeled globally and their local variations among species populations are addressed separately by theoretical and experimental approaches. In this thesis, we introduce a local modeling method to identify differences in species-habitat associations from distribution data. At the example of Douglas fir (Pseudotsuga menziesii), we show how the model coefficients of a spatially varying coefficient model can be clustered in groups. Further, we examine differences between the groups with regard to environmental effects and optimal habitat conditions. Finally,we validate the results with reference genotypes. This approach complements global modeling methods and supports research on impact in climatic change on a species’ distribution and migration.


Bayesian Calibration of a Process Based Forest Model 

Stefan Paul
: Bayesian Calibration of a Process Based Forest Model (2017)
Betreuer: Prof. Dr. Carsten F. Dormann

Forest ecosystems cover 31 percent of the earth’s terrestrial surface and provide a wide range of functions and services relevant to humans. With imminent climate change, it becomes increasingly important to make quantitative predictions about the future of these ecosystems. Process-based models are recognized as state-of-the-art method to predict forest responses to a changing climate. A challenge for working with these models, however, is the correct quantification of parameters and uncertainties from inputs to model parameters and internal processes up to outputs and observation processes. In recent years, Bayesian inference has emerged as a comprehensive framework to estimate parameters and quantify uncertainty in process-based vegetation models. So far, however, the error structures used in these analysis were nearly always limited to observation errors only. This neglects two important factors that are arguably always present in complex vegetationmodels: a) model error b) stochasticity in the internal processes of the model. A potential solution are state-spacemodels. State-spacemodels are hierarchical stochastic models that consider both internal process errors and observational error. Using the state-space framework promises amore realistic quantification of model uncertainties, and might even allow to identify “problematic” regions in the model that are associated with systematic bias. The disadvantage of the method is that the parameter estimation in a state-space framework is substantiallymore costly than in an observation-error only framework. Here, we demonstrate the advantages of nevertheless investing into using the state-space framework for parameter estimation of a vegetation model. We fit a dynamic vegetation model with a range of data, including synthetic data created with process stochasticity and process error, both with a standard observationerror only and a state-space approach. For the state-space estimation, we use a novel ParticleMarkov ChainMonte Carlo sampler. Our results demonstrate that the state-space approach performed better than a standardMarkov ChainMonte Carlo approach for the same scenarios providing correct parameter estimates even in the presence of process errors or model bias. In conclusion the simulated scenarios emphasize the potential of the statespace approach for ecology. Since all models are inevitably imperfect, state-space models cannot only provide better parameter estimates but also allow to quantify the uncertainties in model processes. For further applications a main focus should lie on the development of a specifically designed process-based statespace forest model to fully profit from the advantages of the approach.


Sensitivity analysis of camera trap based animal population estimates 

Peter Antkowiak
: Sensitivity analysis of camera trap based animal population estimates (2017)
Betreuer: Prof. Dr. Carsten F. Dormann

With the development of motion triggered digital cameras that can record almost unlimited amounts of pictures, monitoring wildlife has become easier and cheaper over the past few years. However, most non-invasive camera-based counting schemes for animal populations are either relative density indices or pseudo capture-recapture designs that require recognizing single individuals. In 2008, Rowcliffe et al. proposed and demonstrated the Random Encounter Model that creates density estimates from camera trap data based on the assumptions of an ideal gas model. In addition to the number of photographs, only camera parameters and the average animal speed and group size are required. Since then, the approach has been extended and used by a handful of studies in different settings. While the approach was generally deemed usable, there are some constraints regarding variance and confidence intervals. Common problems identified are the correct estimation of animal speed, measurement of camera parameters, high variance in picture rates and bias induced by non-random camera placement. Due to the broad range of factors that can create biases, applicability has to be proven individually for each species and scenario. Conservation authorities in Germany, primarily in the Bavarian Forest National Park want to use the Random Encounter Model to estimate roe and red deer populations. The practical question in this case is what biases are to be expected and how the cameras should be placed to minimize bias and variance of the density estimates. In the first part of this study I identify and review possible factors affecting bias and variance of and evaluate their expected impact and relevance. The second part deals with a case study: Based on an extensive telemetry data set and detailed habitat models I evaluate whether the Random Encounter Model is suitable for red and roe deer in the Bavarian Forest and determine a robust and efficient sampling protocol. For those movement patterns found to induce bias, I develop methods to correct the estimates. This is achieved by amending the Random Encounter Model with information from habitat models. I further demonstrate that the proposed correction can be used to economically optimize the sampling. For quantifying variances and biases throughout the study I rely on literature as well as on analytical calculations and simulations. My research shows that the Random Encounter Model can deliver reliable density estimates if the sampling effort is sufficiently high and the protocol is followed accurately. Incorrectly measured animal speed and seasonal correlations between speed and density pose a risk for bias. Small scale variance in density is identified as the main driver of variance in the estimates. Detailed habitat models allow for bias correction and an optimized sampling design that reduces the minimum sampling effort by more than 50 percent.


Beschreibung des Sommerhabitats der Westlichen Smaragdeidechse Lacerta bilineata am Kaiserstuhl und Tuniberg 

Martin Hoffmann
: Beschreibung des Sommerhabitats der Westlichen Smaragdeidechse Lacerta bilineata am Kaiserstuhl und Tuniberg (2016): resolveuid/5825e061c4bb6e82e85360fa3225a8f1
Betreuer: Prof. Dr. Carsten F. Dormann

Der menschliche Einfluss auf Natur und Umwelt erreicht seit der industriellen Revolution zuvor nicht gekannte Ausmaße (Tarsitano, 2006). Die Zivilisation und damit auch die Landnutzung nehmen immer mehr Raum ein. Ein Fünftel der Landmasse ist bereits landwirtschaftlich überformt (Hoekstra et al., 2005). Ganze Ökosysteme sind hiervon betroffen (Ribeiro et al., 2009), da oft grundlegende Bestandteile wie die Böden (Dupouey et al., 2002), die Pflanzengesellschaften (Verheyen et al., 2003), sowie die Zusammensetzung der Fauna, sowohl der Invertebraten (Paoletti, 2012) als auch der Vertebraten (Chamberlain und Fuller, 2000; Heikkinen et al., 2004) durch den Menschen stark verändert werden. Die Landnutzung wird als Hauptursache des Verlustes an Biodiversität angesehen (Sala et al., 2000). Das Artensterben erreicht Ausmaße, das Begriffe wie „Massenvernichtung“ (Cushman, 2006) oder „Auslöschungskrise“ (Hoekstra et al., 2005) verwendet werden. Global gesehen ist die Aussterberate von Pflanzen und Tieren bis zu 1000 Mal höher als eine aus Fossilienfunden ermittelte Vergleichsrate (Wilson, 2010; Baillie et al., 2004). Amphibien und Reptilien sind von allen terrestrischen Vertebraten am stärksten betroffen, mit mehr gefährdeten Arten als bei Vögeln oder Säugetieren (Gardner et al., 2007; Baillie et al., 2004). Als Hauptursachen der Gefährdung von Amphibien und Reptilien werden invasive Arten, Umweltverschmutzung, Krankheiten, Klimawandel, Habitatdegradation und Habitatverlust genannt (Gibbons et al., 2000). Habitatverluste führen häufig zur Zerschneidung von Populationen. Durch Fragmentierung und Isolation nimmt das Risiko einer Auslöschung einer Population zu (Burkey, 1989).


Ecosystem services in agroforestry systems of Europe 


Amelie Göbel: Ecosystem services in agroforestry systems (2016): resolveuid/5825e061c4bb6e82e85360fa3225a8f1
Betreuerin: Anne-Christine Mupepele

Agroforestry systems are de€ning elements of the European countryside [McAdam and McEvoy, 2009] that are viewed as part of a working landscape and provide ecosystem services, environmental bene€ts and economic commodities [Jose, 2009] [Bj¨orklund et al., 2013]. However, agroforestry systems in Europe are not very well explored in terms of ecosystem services and their evidence basis [Fagerholm et al., 2016]. Œe current state of knowledge, as well as the gaps in the research into ecosystem services in agroforestry systems in Europe, will be identi€ed and a subset of studies will be critically appraised, centred upon the quality of the literature. A systematic literature map, following the guidelines established by the Collaboration for Environmental evidence (CEE) Collaboration for Environmental Evidence [2009] was therefore conducted. Œe ISI web of knowledge database and the search engines, such as Google and Google scholar, were systematically surveyed for relevant studies on the topic. Œese searches were limited to English and German articles, published since 1929. Œe search results were assessed for relevance in a two step process of comparing title and abstract, as well as abstract and the full text against stipulated and/or designated criteria for both inclusion and exclusion. Œe remaining 110 articles were then qualitatively evaluated using the so‰ware R. As a second step, the quality of a subset of the €nal articles was critically appraised using the quality assessment tool established by Mupepele et al. [2015]. We found amongst the 110 studies included that regulation maintenance and provisioning services are the most investigated ecosystem services categorize in the literature. A major part of the literature focuses on dehesa and silvopastoral systems located mainly in the Mediterranean region. Œe Nemorales zonobiome (temperate climate) have not so far been researched extensively regarding ecosystem services in agroforestry systems. Œe €ndings also show distributions of the the studies concerning the study design and the main message and/or implications of the articles. Œe evidence assessment generally revealed that many studies remained either at the same level of evidence or are downgraded only to approximately 0.5 of a level. To conclude, this work can be seen as the point of commencement in ecosystem services for agroforestry systems research in Europe. It can be used by researchers, practitioners, conservationists and policy-makers as reference for prospective agroforestry projects.


Resource selection and lƒight activity of Leisler’s bats (Nyctalus leisleri) in Gundel€ngerWald during mating season 

Laura Kehry
: Resource selection and lƒight activity of Leisler’s bats (Nyctalus leisleri) in Gundel€finger Wald during mating season (2016)
Betreuer: Dr. Simone Ciuti

Sex-speci€fic behaviour as a result of asymmetric reproductive strategies is a common phenomenon in mammals. Most temperate bats, including Leisler’s Bats (Nyctalus leisleri), are sexually segregated during breeding season in summer. Sex-speci€c behavioural pattŠerns are well studied during this reproductive phase. During mating season in autumn, both sexes share resources such as roost sites. However, littŠle is known about their nightly behaviour during this time of the year. Œis study investigated resource selection and lƒight activity paŠtterns of N. leisleri during mating season by using radio telemetry. ThŒree females and four males were radio tracked for several nights each. Space use was analysed with a resource selection function (RSF) including habitat, hydrological, human disturbance and topographical variables. Flight activity during the course of the night was modelled with a generalised additive model (GAM) taking into account weather conditions. Sex-specifi€c eff‚ects were found for most of the spatial parameters examined. Home ranges and maximum distances from roost sites were two to three times larger for females compared to males. Œe overall eff‚ects of habitat, distance to urban area, distance to forest edge, slope and aspect on resource selection were sex-speci€c, while the e‚ffects of distance to day roost and distance to major roads di‚ffered only in eff‚ect size. For males, small distance to their roost sites and ƒat terrain were the main drivers of resource selection, while for females, all variables were of equal importance. Moreover, lƒight activity pattŠerns throughout the night diff‚ered between sexes. Of the weather variables investigated, temperature had the strongest inƒfluence on lƒight activity, even though wind speed, precipitation and wind direction had a signifi€cant e‚ffect too. ŒThese results indicate that, besides foraging behaviour, courtship behaviour of males has a strong inflƒuence of spatial and temporal activity pattŠerns of Leisler’s Bats during mating season. Even though females and males share roosts at least for short times, resource partitioning occurs at night. ThŒe €findings of this study underline the seasonal variations in sex-specifi€c resource requirements of bats. ŒThese are important to know in order to design eff‚ective conservation strategies for entire populations.


Analyzing the predictive performance of LiDAR-derived metrics in modeling fine-scale resource selection by deer 

Henriette Tripke
: Analyzing the predictive performance of LiDAR-derived metrics in modeling fine-scale resource selection by deer (2016)
Betreuer: Dr. Simone Ciuti

To disentangle the complexity of wildlife responses to their environment, animal behavior and their habitat requirements have to be understood on an increasingly fine scale. Remote sensing provides advanced techniques to describe habitat characteristics used by animals.In particular light detection and ranging (LiDAR) is increasingly used to describe threedimensional vegetation structure, occasionally in extremely high resolution. Due to its successful application in forest inventory LiDAR data have become more and more available to wildlife ecologists and showed already great potential for understanding fine-scale behavioral patterns in animals.In this study I tested the predictive performance of high resolution LiDAR data in combination with fine-scale telemetry data. I implemented a resource selection function (RSF) for two deer species native to the Bavarian Forest National Park. In a systematic model design using conditional logistic regressions I compared the predictive power of different LiDAR-derived metrics to more commonly used remote sensing data. Moreover, I applied a newly developed LiDAR product, which was expected to capture vegetation structure better than other common LiDAR products. A principle component analysis (PCA) was used to generate new, linearly uncorrelated variables, so-called principle components (PCs), from the total LiDAR point cloud.


Evaluation of Aerial Blimp and Visual Boat Surveys for Counting the Araguaian River Dolphin (Inia araguaiaensis) in the Cantão State Park, Brazil 

Julia Sophie Fürstenau Oliveira
: Evaluation of Aerial Blimp and Visual Boat Surveys for Counting the Araguaian River Dolphin (Inia araguaiaensis) in the Cantão State Park, Brazil (2015)
Dr. Simone Ciuti

A blimp (non-rigid airship system) mounted with a camera was used to compare an aerial survey method with a visual boat survey, counting the recently described Araguaian River dolphin (Inia araguaiaensis). The study was carried out in the low water season (August – November 2015) in the Cantão State Park (PEC), a seasonally flooded lake system in the transition zone of Amazonian forest and Savanna in the State of Tocantins, Brazil. I tested effects of survey methods on dolphin sighting, including factors that influenced dolphin abundance and sighting. Six sectors (4.27 km2) were replicated, applying full counts in strip transects. A boat with three observers travelled mid-river and a second boat in which the pilot held the blimp, was navigated simultaneously. The camera recorded videos continuously. The aerial survey method detected significantly more river dolphins. The observers sighted 74.1% of the dolphins sighted by the blimp method. In wider sections, significantly larger group size was encountered by the blimp than by the visual survey method. More adult dolphins and calves were encountered by both survey methods in sections of larger area. Regarding habitat, the highest dolphin abundance was found in mouths of arms and lakes. A high dolphin density was found in the surveyed area (19.7/km2). To enhance river dolphin conservation, robust abundance and distribution data are needed. With further development of blimps and also drones, aerial methods are promising for providing inexpensive and efficient surveys.


Frequency dependent pollination for rewarding species in a combined approach of fi€eld data and agent-based modeling 

Helen Czioska
: Frequency dependent pollination for rewarding species in a combined approach of €field data and agent-based modeling (2015)
Dr. Gita Benadi

Flower frequency dependence occurs if the frequencies in the flƒowering community a‚ffect the foraging behavior of pollinators. By inƒfluencing the plants €fitness though di‚ffering visitation rates, frequency dependent pollination could have far-reaching consequences for plant coexistence. Negative frequency dependence, hence a pollinators preference for the rare species, is thought to enhance diversity and to be the reason for color morphisms in rewardless orchids. Common species on the other hand bene€fit from a positive frequency dependence which can reduce diversity. However, only few studies have been conducted on frequency dependence and are inconsistent their results. Focus in this thesis is the analysis of frequency dependence for rewarding species in natural ƒflower communities and the identifi€cation of inƒfluencing factors using a combined approach of €field and model data. I observed pollinator visitation to ƒflower of fi€ve rewarding species in their natural plant community in the area of the Jena Experiment. ŒThereupon I used an agent-based model of two co-ƒflowering plant species competing over pollination service by a shared pollinator to identify paŠtterns and in- ƒfluencing factors of frequency dependence. Four out of fi€ve species showed a cubic frequency dependence in the fi€eld data. Furthermore, the results of the model support the general relationship and identify ƒoral cover, cluster size and reward as important inƒfluencing factors. Negative frequency dependence was found for high ƒoral cover, positive for low cover or very high reward. In conclusion, my results indicate the existence of frequency dependent pollination for rewarding species in a natural flƒower community. Also, frequency dependence appears to depend on ƒoral cover and spatial aggregation of flƒowers. ThŒerefore, paŠtterns of frequency dependence are likely to change in time and space and are not solely related to reward and certain species. In consequence,, frequency dependence might be more important concept than previously thought for evolution and maintenance of diversity.


Vigilance behaviour in Alpine ibex (Capra ibex) and Alpine chamois (Rupicapra rupicapra) kids 

L. Hannah C. Weber
: Vigilance behaviour in Alpine ibex (Capra ibex) and Alpine chamois (Rupicapra rupicapra) kids (2015)
Dr. Simone Ciuti

In prey species, vigilance is an important behavioural trait with regard to antipredator strategies. It has been studied intensely in animals. However, liŠttle is known about vigilance in animal kids. In this study, I investigated the inƒfluence of various environmental and social factors on Alpine ibex (Capra ibex) and Alpine chamois (Rupicapra rupicapra) kid vigilance behaviour. In the Gran Paradiso National Park, Italy, animals were observed from June to August, 2015, using group scan and focal kid observations. Data was analyzed using general additive and general additive mixed models as well as a cross-correlation function. Œe results revealed that there is no interspeci€c diff‚erence in kid scanning behaviour. A group size e‚ffect on kid vigilance is given, however no group size e‚ffect could be found for adults. Kid vigilance is stronlgy aff‚ected by the distance to the closest hiking trail, which can be seen as a proxy for human disturbance. Œis e‚ffect does not exist for adults, which suggests a habituation process. Against my expectation, the fear of the group is not reƒflected in kid scanning behaviour. In general, kids are more vigilant than adults. Animals scan more in rocky terrain than in grassland and more in the morning than in the evening. Kids synchronize their scanning behaviour with that of adults, which suggests a learning e‚ffect between generations. Œe innovative focus of this study on vigilance behaviour in kids reveals that there are still a lot of open questions in this fi€eld.



Can we identify community assembly from phylogenies? A test with a biogeographic simulation for predicting phylogenetic community structure and environmental niche evolution 

Jan Paul Bauche
: Can we identify community assembly from phylogenies? A test with a biogeographic simulation for predicting phylogenetic community structure and environmental niche evolution (2015): resolveuid/3c03a85704f7af02248fe6a922eb6237

Community assembly processes shape species communities in terms of biogeographical distribution and local coexistence. Such processes can be for example environmental niche adaptation which leads to habitat €ltering or local competition resulting from inter- and intra-species similarities. Such similarities would, following the idea of limiting similarity, lead to the concurrence of less similar species. The question which mechanisms are responsible for the assembly of a specific community and how they can be identified is a long discussed topic in evolutionary ecology and has recently received renewed attention due to new types of data becoming available. The idea of identifying these mechanisms via the phylogenetic structure of a community has moved into focus. However it remains unclear and strongly debated how reliable inferences of community assembly processes based on phylogenetic data actually are. In this study I produce patterns of phylogenetic clustering and spatial species distribution, derived from simulations of community assembly processes to identify the resulting combinations. This allows me to assess the suitability of phylogenetic clustering or overdispersion as an indicator for community assembly processes and to show which patterns are best suited for the identification of the mechanisms which lead to the assembly of a species community. I found that phylogenetic clustering alone does not provide sufficient information to identify the underlying processes. A combination with spatial patterns such as global and local rank-abundance curves does not result in a clearer picture. I show that the combination of phylogenetic and spatial patterns of different scales is not sufficient to predict species community assembly processes.


Understanding dispersal ecology of young male ungulates (Cervus elaphus) to design large scale wildlife corridors 


Robin Benz: Understanding dispersal ecology of young male ungulates (Cervus elaphus) to design large scale wildlife corridors (2014)
Dr. Simone Ciuti

Human modifications within natural habitats, such as linear clearings and roads, fragment the landscape into ever smaller habitat patches and reduce the permeability to wildlife movement. Habitat modifications have occurred exponentially over the last decades, with global biodiversity declining at the level of genes, species and ecosystems, while the majority of pressures showing increasing trends. Thus, conservation efforts focus on reservation and maintenance of wildlife corridors. In this study, I investigate the movement ecology of young male ungulates Cervus elaphus in order to design a large scale corridor matrix across the Northern Rocky Mountains. I used high-resolution satellite telemetry data of 54 young male individuals monitored from 2007 - 2011 to combine broad-scale habitat preference with fine-scale movement behavior. The majority of montane elk are migratory, meaning individuals move from wintering areas to summer ranges at higher elevations, with the exception of young males, which commonly disperse or make further exploratory movements. I used resource selection functions (RSFs) to identify best winter and summer hotspots selected by elk. For elk movement I separated long directional steps from short resting and foraging ones, and predicted fine-scale habitat selection using step selection functions (SSFs), while considering spatio-temporal differences in movement based on extrinsic factors, such as seasonality for spring and hunting in autumn. Elk showed marked habitat selection during both stationary and exploratory periods, and strong avoidance of roads, in particular high traffic highways. Based on SSF predictions, I build two cost friction surfaces for each season, where elk (i) moves on natural behavioral choice, and (ii) where elk is constrained by roads. Based on these, I mapped least cost corridors (LCCs) between winter and summer hotspots. Quantitative comparison of highway-corridor intersection lengths revealed significant effect of major highways on elk connectivity. This analytical tool can be used to define hotspots of elk connectivity along major roads, and can be easily applied to other large mammals and primarily aid conservation managers to make better informed management decisions based on animal movement behavior.


Hotspots of fear: Effects of wolves on tree recruitment in the Białowieża Primeval Forest 


Hannah Heither: Hotspots of fear: Effects of wolves on tree recruitment in the Białowieża Primeval Forest (2014)
Betreuer: Dr. Simone Ciuti

The landscape of fear concept has been extensively studied in different large scale ecosystems such as the North American Yellowstone National Park. In this area, elk (Cervus elaphus) have been shown to change large scale movement and foraging patterns to avoid predation by wolves (Canis lupus). After reintroduction of wolves and consequent effects on elk occurrence and distribution, several tree species have recovered due to decreased elk browsing pressure. In an European temperate forest system, the Białowieża Primeval Forest, a recent study suggests small scale shifts (<1km) of ungulate behaviour caused by the presence of wolf. Woodland habitat characteristics such as coarse woody debris (CWD) were considered to act as escape impediments for red deer (C. elaphus) that resulted in reduced local browsing pressure. However, the effect of this lower browsing pressure on tree recruitment in the vicinity of CWD has not yet been verified. In this study, I compared tree seedling numbers and tree recruitment effects of presence of CWD combined with a wolf core area and a gradient of wolf activity in the strict reserve in the Białowieża Primeval Forest. I sampled seedling number, seedling species and seedling height (<2m) in plots with CWD paired with control plots without CWD in areas of deciduous forest. Additionally, I recorded features of CWD (e.g. height, length, presence of branches), as well as potential confounding factors such as the canopy openness. Successful recruitment of a tree was defined as seedling tree height exceeding ungulate browsing reach (>2m). The data I gathered required to be controlled for zero truncation, overdispersion, and zero inflation issues, and eventually were modeled using zero truncated generalized models and negative binomial mixed effects zero inflated models. In regard to the occurrence of seedlings, their numbers were not facilitated by the presence of CWD neither by wolf occurrence. Indeed, I only found a quadratic effect of canopy openness on seedling occurrence, with an optimum at 20% of canopy closure. However, tree recruitment (e.g., increased likelihood of seedlings reaching 2 meters of height) was facilitated in direct vicinity of CWD (Kaplan Meier Survival Analysis, ρ1<0.0001). Within control plots (no-CWD), tree recruitment was significant lower due to higher mortality or overbrowsing pressure, leading plants to remain short. Interestingly, higher numbers of seedlings were associated with increasing CWD height (p=0.048), irrespective to the probability of encountering the wolf. Other characteristics (presence of branches, uprooted tree, CWD length, CWD species, CWD age class and CWD volume) were not significantly contributing to the model. This study is the first one that indicated an effect of a certain CWD feature on tree recruitment. My results included a first comprehension of the underlying behavioral mechanism. Red deer avoided browsing close to high CWD, which is likely, a result of reduced visibility and higher escape impediments. The avoidance of this habitat characteristic appears important for tree recruitment close to CWD in this small scale European forest situation. High CWD seems perceived by red deer as a ‘hotspot of fear’ throughout the whole area of the strict reserve.


Analysis of the natural tree mortality in windthrow areas in the Bavarian Forest National Park, Germany 


Gregor Didenko: Analysis of the natural tree mortality in windthrow areas in the Bavarian Forest National Park, Germany (2012): resolveuid/66b04045a3ca5f454c3a32897be685f6 

Understanding and analyzing the highly stochastic process of natural tree mortality is one of the hardest tasks of forest growth modeling. The mortality splits into two parts: regular and irregular mortality. In this study regular mortality was analyzed using individual tree data from permanent monitoring plots in the Bavarian Forest National Park, Germany. The data incorporated up to 20 years of periodical re-measurements. The plots were established in the late 1980s to follow the natural reforestation on wind-thrown stands. Six sites were chosen on three different altitudes. On each altitude one site was cleared of all the dead wood and uprooted trees while another site was left uncleared, in its current state. Using the Kaplan-Meier-Estimator and generalized mixed effect modeling the effects of the explanatory variables on individual tree mortality were analyzed and quantified. The explanatory variables included factors that potentially influenced the likelihood of mortality and were available in the data: tree height, site treatment, individual tree competition and browsing. Competition was assessed with Hegyi’s distance dependent competition index. This index explicitly takes the distance between subject tree and neighbor into account as well as tree specific vigor values, in my case the tree height. The parameter estimates of all used predictors had been found highly significant (P < 0.001) in influencing tree mortality. The results further indicated that the effect of height on mortality differs in between high and low competitive environments. Browsing and the clearing of the sites increased chances of mortality by over 50%. Significant differences were found between the tree survival curves on different treatments as well as between high and low competition index values. My study introduced the first application of tree diameter-classes as surrogates for the time-variable in the survival analysis. It provided detailed insights not only into the probability of growing into a given DBH-class but also to the total growth rate of different risk groups. The results of my study are expected to be incorporated into a forest growth model of the Bavarian Forest National Park. Analyzing tree mortality on wind throws may even contribute to the contradictory discussion on how to treat wind thrown forests. 


— filed under:
Personal tools