Data Analysis in Vegetation Ecology

Otto Wildi

• Editorial: CAB International
• Año: 2017
• ISBN:
  9781786394248
• Edición: 3
• Categoría: Biowissenschaften
• Categoría: Ökologie

The 3rd edition of this popular textbook introduces the reader to the investigation of vegetation systems with an emphasis on data analysis. The book succinctly illustrates the various paths leading to high quality data suitable for pattern recognition, pattern testing, static and dynamic modelling and model testing including spatial and temporal aspects of ecosystems. Step-by-step introductions using small examples lead to more demanding approaches illustrated by real world examples aimed at explaining interpretations. All data sets and examples described in the book are available online and are written using the freely available statistical package R. This book will be of particular value to beginning graduate students and postdoctoral researchers of vegetation ecology, ecological data analysis, and ecological modelling, and experienced researchers needing a guide to new methods. A completely revised and updated edition of this popular introduction to data analysis in vegetation ecology. Includes practical step-by-step examples using the freely available statistical package R. Complex concepts and operations are explained using clear illustrations and case studies relating to real world phenomena. Emphasizes method selection rather than just giving a set of recipes.

1 Introduction

Irrespective of whether our interest in the vegetation of the earth evolves from naive curiosity or from the endeavour to conserve this natural resource, the study of vegetation ecology inevitably follows the rules of epistemology and it is shaped by paradigms.

1.1 Epistemology

This book is about understanding vegetation systems in a scientific context, a topic of vegetation ecology. It is written for researchers motivated by the curiosity and ambition to assess and understand vegetation dynamics. Vegetation, according to van der Maarel and Franklin (2013) ‘can be loosely defined as a system of largely spontaneously growing plants’. What humans grow in gardens and fields is hence excluded. The fascination of investigating vegetation resides in the mystery of what plants ‘have in mind’ when populating the world. The goal of all efforts in plant ecology, as in other fields of science, is to learn more about the rules governing natural systems. These rules are causing patterns, and the assessment of patterns is the recurrent theme of this book. Albeit sometimes patterns are striking, their proper recognition can be elusive for various reasons.

We are aware that access of humans to the real world is rather restricted and – as we know from experience – differs among individuals. Our sins provide us with a limited set of signals and even this is far too voluminous to be comprehensively processed by our brain. We usually implicitly accept that the complexity of this world exceeds our imagination. Furthermore, the extents of our target systems are frequently excessive. For example, the data set addressed in Chapter 13 uses data collected within plots of 200 square metres, yet spaced by as much as 4 kilometres. And finally, humans are part of the natural system and not independent observers as would be needed for truly unbiased recognition.

To achieve progress in research an image of the real world is needed: the data world. In this we get a description – an image – of the real world in the form of numbers. (An image can be a spreadsheet filled with numbers, a digital photograph, a digital terrain model, or other.) Assuring that such an image reflects properties of the system is a challenge – it is the task of sampling design, as yet under-represented in the literature of vegetation ecology. The point in mind is to avoid personal bias, resulting in conclusions triggered by the investigator. All examples in this book result from attempts to avoid bias, except when mentioned explicitly. Data also deserve various kinds of treatment, like data management, quality control and of course analysis. But what constitutes the final issue of data analysis?

Upon analysis we develop our model world describing our understanding of the real world. This understanding is nothing else but a hypothesis about the state and functioning of reality. The model world too is often represented in the form of numbers, although typically less voluminous and complex than the numerical description of the real world. It can therefore also be the subject of further analysis. An advanced issue of data analysis is in relating the data world to the model world and vice versa. Finding globally valid models reflecting the real world is a difficult task due to the complexity of systems (Orlóci 1993; Anand 1997). Complexity has its origin in some fairly well-known phenomena, one being the scale effect. Any pattern in ecosystems will emerge at a specific spatial and temporal scale only: at short spatial distance competition and facilitation among plants can be detected (Connell and Slatyer 1977); these would remain undetected over a range of kilometres. In order to study the effect of global climate change (Orlóci 2001; Walther et al. 2002) the scale revealed by satellite imagery is probably more promising than the same of a local survey. Choosing the best scale for an investigation is a matter of decision, experience and often exploration. For this a multi-step approach is needed, in which intermediate results are used to evaluate the next decision in the analysis. Poore (1955, 1962) called this successive approximation, Wildi and Orlóci (1991) flexible analysis and Albert et al. (2010) model-based sampling. Hence, the diversity and flexibility of methods is nothing else but a response to the complex nature of systems.

Due to complexity there exist alternatives of valid models, for instance, amalgamating different spatial scales. And once a proper scale is found there is still a need to simultaneously consider an ‘upper’ and a ‘lower’ level, because knowing sensitivity to change in scale is essential. Parker and Pickett (1998) discuss this in the context of temporal scales and interpret the interaction as follows: ‘The middle level represents the scale of investigation, and processes of slower rate act as the context and processes of faster rates reflect the mechanisms, initial conditions or variance.’

Another source of complexity is uncertainty in data. Data quality often suffers from practical constraints, for example from limitation in time, money and accessibility (Albert et al. 2010). A detailed vegetation survey is time-consuming, and while sampling, change may already be underway (Wildi et al. 2004). Such data will therefore exhibit an undesired temporal trend. A specific bias causes variable selection. For example, it is easier to measure components above ground than below ground (van der Maarel and Franklin 2013, p. 6), a distinction vital in vegetation ecology. Once the measurements are complete they may reflect random fluctuation or chaotic behaviour (Kienast et al. 2007) blurring deterministic components. It is a main objective in data analysis to distinguish random from deterministic, either linear or nonlinear effects. Nonlinearity would not be a problem if we knew the kind of relationship that was hidden in the data (e.g. Gaussian, exponential, logarithmic; Austin 1987), but finding a proper function is usually a non-trivial problem.

Spatial and temporal interactions add much to the complexity of vegetation systems. In space, the problem of direction arises, as the order of objects depends on the direction considered. In most ecosystems, the environmental conditions, for example elevation or humidity, change across the area. Biological variables responding to this will likely behave similarly and therefore become space-dependent (Legendre and Legendre 2012). Even if there is no general trend in space, a local phenomenon may exist: spatial autocorrelation. This means that sampling units in close neighbourhood are more similar than one could expect from ecological conditions. Similarly, correlation over time exists as well. In analogy to space, there occur temporal dependence and temporal autocorrelation. Many processes are temporally continuous and the systems will usually change gradually only, causing two subsequent states to be similar. Finally, time and space are not independent, but linked. Spatial patterns too tend to change continuously over time. Therefore, a time series observed at one point in space is expected be similar to another series observed nearby.

In summary, all knowledge we generate by analysing the data world contributes to our model world which in the end is aimed at serving society. When translating this into practice we meet yet another world, the man-made world of values. This is people’s perception and valuation of the world, which we know from experience is continuously changing. The results we derive in the course of analysis carry the potential to deliver input into value systems, but we should keep in mind what Diamond (1999) mentioned when talking about accepting innovations: ‘Society accepts the solution if it is compatible with the society’s values and other technologies.’ Assessing the existence of global warming, as an example, can be a matter of modelling. Convincing people of the practical relevance of the problem is a question of evaluation and communication, skills not addressed in this book.

1.2 Paradigms ruling analysis

To date a comprehensive and proprietary theory of the plant community is still missing (McGill et al. 2006; Feoli and Orlóci 2011). However, in the course of the history of data analysis in vegetation science (which may have had a starting point with the invention of a similarity coefficient by Jaccard in the year 1901) various theories have been inherited from other fields of science and new paradigms evolved and became widely accepted. As can be seen below these are usually related, sometimes depicting the same idea from a different point of view. They contribute considerably to our present understanding of the functioning of plant-environment systems.

Similarity theory. Feoli and Orlóci (2011) recently emphasized the universal role of ‘similarity theory’ in the analysis of multivariate biological data. This theory has equal importance in other fields of science like genetics or meteorology, and it encompasses various axioms, as for instance:

1.   The similarity of any two objects (plots, for instance) described by many variables (species, chemical or physical measurements, etc.) can be measured or estimated.

2.   Similarity applied to entire sets of objects (addressed as samples in Section 2.3) constitutes the similarity pattern of these sets (see Section 4.5).

3.   Similarity (including similarity patterns) ‘can express linkages to factors which control the existence of the objects’ (Feoli and Orlóci 2011). This relies on the assumption that any two objects that are similar in terms of vegetation are expected to be similar in terms of growth conditions as well, and vice versa.

Similarity theory, according to Feoli and Orlóci (2011), plays a crucial role in the joint analysis of biological, environmental, spatial and temporal data. To this end it also yields the justification to research on similarity functions (Chapter 4). Needless to say that the majority of methods described in this book rely on similarity or its complement, distance. But similarity relates to various other paradigms as well.

Convergent evolution. This encapsulates the observations that ‘plants of different species develop similar morphology to adapt to similar environment’ (Feoli and Orlóci (2011), a fact discussed in many papers, such as Orlóci and Orlóci (1985), Feoli and Orlóci (1991), Díaz (1995), Pillar (1999), Tobias et al. 2014). A famous example is the succulent life form of the Cactaceae in the New World and its counterpart of Euphorbiaceae in the Old World (Wildi and Orlóci 2007). Convergent evolution justifies the use of characters other than taxonomic, that is, genetic, morphologic or functional (Chapter 8), for example. As a consequence of convergent evolution the principles of similarity theory equally apply to plant traits used for vegetation description.

Convergence and divergence. According to similarity theory similar plant communities correspond to similar physical-chemical environments (see above). Therefore, in the course of vegetation change, patterns of plant communities and environmental factors tend to converge (that is, tend to get more similar) and convergence may become an element of our model world (MacArthur and Levins 1967; Pillar et al. 2009). Convergence is sometimes used in a static context, simply expressing that patterns in plant communities and the environment are similar. But there are also opposing forces causing divergence. Species requiring similar environmental conditions for survival are competing for the same ecological niche. Hence, the least fit may be expelled by competition. Any final plant composition is therefore the result of convergence and divergence acting simultaneously.

Vegetation dynamics. Vegetation change too follows similarity rules and it is therefore potentially predictable. Notwithstanding environmental conditions, Feoli and Scoppola (1980) hypothesize that ‘a given plant association has the highest probability of transition to the one with which it shares maximal similarity’. This principle also explains traditional succession which is therefore just a special case of vegetation dynamics. And, as mentioned by Feoli and Orlóci (2011), it is also applicable in the opposite direction, when disturbance counteracts convergence, for example.

Filtering. Filtering is a popular term explaining the constrained set of species or species traits in a stable plant community (Keddy 1992; Fridley 2003). The principle is sketched in Figure 1.1 with three different filters modulating four species pools. Such a filter, according to Keddy (1992), does nothing but natural selection on the community level. Filtering invokes some general principles in the formation of plant communities:

Figure 1.1: Principles of the filter concept. A species or traits pool is reduced stepwise by different kinds of processes called filtering.

1.   The formation of any plant community is strictly limited by the pool of species or the pool of traits.

2.   If a pool is changed, for instance by local evolution, this will directly translate into the final composition of plant communities. This aspect is not included in the scheme of Figure 1.1.

3.   The idea of filtering perfectly describes what can happen when alien species (‘neophytes’) invade. First, invasion extends the species pool and second, it affects biological filters (competition, for example) to eventually generate entirely new combinations of species and traits.

Several of these principles are sometimes subsumed under the term ‘assembly rules’ despite the fact that filtering is a selecting rather than a combining process. But as in all paradigms mentioned so far, it is another nice outset for generating hypotheses. It is the task of analysis to accept or reject the latter using methods of data description and statistics.

2 Patterns in vegetation ecology

Patterns are recurring regularities, and pattern recognition is a target of data analysis in vegetation ecology. A proper sampling design is required to deliver data reflecting properties of the real system. Similarity patterns eventually occur in any sub-system, in resemblance space, in geographical space and in time.

2.1 Pattern recognition

Why search for patterns in vegetation ecology? Because the spatial and temporal distribution of species is nonrandom. The species behaviour is governed by rules causing detectable, recurring patterns that can be described mathematically, such as by a straight line (a regression line, for example) or ...

Índice

• Cover Page
• Title Page
• Copyright Page
• Dedication
• Contents
• List of Figures
• List of Tables
• Preface to the first edition
• Preface to the second edition
• Preface to the third edition
• 1 Introduction
• 2 Patterns in vegetation ecology
• 3 Transformation
• 4 Multivariate comparison
• 5 Classification
• 6 Ordination
• 7 Ecological patterns
• 8 Traits and indicators
• 9 Static predictive modelling
• 10 Vegetation change in time
• 11 Dynamic modelling
• 12 Revising classifications
• 13 Swiss forests: a case study
• 14 Back to the roots?
• Bibliography
• Appendix A: Functions in package dave
• Appendix B: Data sets used
• Index
• Backcover