Population and metapopulation biology of perennial plants in fragmented grasslands

Zuzana Münzbergová

Doctoral thesis
Department of Botany
Faculty of Science
Charles University
Prague
2004

Index

Introduction
I. Münzbergová Z. and Ehrlén J.: Simpler is better - data collection for demographic studies. Submitted to Oecologia.
II. Münzbergová Z.: Population growth rates of rare and common Cirsium species sharing the same habitat. Submitted to Conservation Biology.
III. Münzbergová Z. and Plačková I.: Long-term population growth rate in populations differing in size and genetic diversity. Submitted to Journal of Applied Ecology.
IV. Münzbergová Z. and Herben T.: Seed, dispersal, microsite, habitat and recruitment limitation - identification of terms and concepts in studies of limitations. Submitted to Oecologia.
V. Münzbergová Z. (in press): Effect of spatial scale on factors limiting species distributions in dry grassland fragments. Journal of Ecology.
VI. Münzbergová Z. and Herben T. (2004): Identification of suitable unoccupied habitats in metapopulation studies using co-occurrence of species. Oikos 105: 408-414.
VII. Münzbergová Z., Mildén M., Ehrlén J. and Herben T. Population viability and reintroduction strategies: a spatially explicit landscape-level approach. Submitted to Ecological Applications.
Conclusions

Introduction

Factors affecting survival of species in fragmented landscapes belong among the central issues in species ecology and conservation biology (van Groenendael et al. 1998). Traditional studies on this issue focus on studying single traits of species or dynamics of local populations (Morris and Doak 2002). With the recent development of metapopulation theory (Hanski 1989), it has been recognized that regional scale processes may also be important for survival of species in the landscape (e.g. Carroll et al. 2003, du Toit et al. 2004). In such cases studying the viability of local populations only is of little use.
Recently several theoretical models have been developed to assess the prospects for species survival at the landscape level (e.g. Gustafson and Gardner 1996, With et al. 1997, Hanski and Ovaskainen 2000, Casagrandi and Gatto 2002a, Dreschler et al. 2003). Most of the models assume that the current distribution of species in habitat fragments represents equilibrium between local colonization and extinction. Under this assumption, data on local population dynamics are not necessary and the pattern of distribution of the species in the landscape alone can be used to estimate the prospects for landscape level survival of the species (Hanski and Ovaskainen 2000).
The critical point in this reasoning is the assumption that colonization/extinction dynamics operate fast enough to keep the current distribution close to equilibrium and that local population dynamics is thus unimportant. Whereas this assumption is likely to be correct in short-lived highly dispersible organisms such as many insects (Hanski et al. 1994, Baquette 2003, Purse et al. 2003), it is less likely to apply in sessile organisms such as plants that are known to have very slow local population dynamics with low extinction and immigration probabilities (Eriksson 1996, Freckleton and Watkinson 2002). Therefore models based on the assumption that species are in equilibrium and that local population dynamics are thus unimportant are of little use in plants. Unfortunately, plants (precisely due to their limited migration capacities) are directly threatened by habitat destruction and predictions of the effects of fragmentation are needed (Eriksson and Kiviniemi 1999). Applying metapopulation theory to plants thus essentially means taking into account processes operating on the local scale and integrating them with processes occurring at the regional scale.
At the local scale species dynamics are determined by the behaviors of individuals. Individuals are however not independent and are influenced by their interaction with other conspecific individuals as well as with other species and the environment. Describing this thus essentially includes studying a range of processes from density dependence in seedling germination to patterns of pollination within the community. Already at this scale population dynamics are affected by landscape changes. The factors include both changes in habitat quality as well as the changes in population size. Habitat quality may affect performance of single individuals as well as their interaction (e.g. Eisto et al. 2000; Vergeer et al. 2003). Population size has important consequences mainly for patterns of mating via attraction of pollinators and by providing a source of pollen, leading to reduced genetic diversity, and consequently to lower fitness (e.g. Mustajarvi et al. 2001; Lienert et al. 2002; Paschke et al. 2002). These direct negative effects are further enhanced by genetic drift and population stochasticity, which play an important role in dynamics of small populations (Ellstrand & Elam 1993; Young et al. 1996).
At the regional scale species dynamics are affected by all the local scale processes as well as by a range of other processes related to both attributes of the species, such as their dispersal ability, and attributes of the landscape, such as the spatial arrangement of suitable habitats and their size. Studies at this scale thus necessarily include studying factors determining habitat suitability for the species, the dispersal ability of the species in the landscape and combine them with information on local population dynamics.
In this thesis I present a series of papers aimed at understanding the population biology and survival potential of perennial plants in fragmented grasslands. In the first paper I explore methodological issues related to demography of local populations. Studies of local population dynamics usually rely on following the performance of single individuals over multiple years and use it to predict the performance of the whole population. This is done by means of matrix population models, which allow combining information from many single individuals to predict their overall behavior. In recent years matrix population models have become a standard method to assess the viability of structured populations (Morris and Doak 2002). Repeated iterations of a matrix result in a projection of a population's equilibrium growth rate that provides a measure of the overall performance of populations. Moreover, sensitivity or elasticity analysis of matrix models can identify the life history stages most critical for the persistence of a species. There is a large literature concerned with the techniques to analyze collected data to make projections or predictions (e.g. Horvitz et al. 1997, Ehrlén and van Groenendael 1998, Ludwig 1999, Mills et al. 1999, de Kroon et al. 2000, Caswell 2001, Ehrlén et al. 2001, Calder et al. 2003). In contrast, very little attention has been paid to how the primary data are actually sampled. This is surprising because the quality of the resulting matrices is primarily a function of how the data were collected and no post-collection manipulations can overcome this.
Almost all recent plant demographic studies have sampled all individuals within plots (Münzbergová, unpubl. data). The most likely explanation for this is that sampling within plots is easy to apply in the field and also provides information on the actual stage distribution. However, for any given sampling effort this strategy is unlikely to provide the most accurate estimates of demographic parameters, since it often results in a very unequal number of individuals per stage. In the literature it is, in fact, not rare to encounter transition probabilities that are estimated using only one or a few individuals (Münzbergová, unpubl. data). Any sampling strategy that is able to decrease the problems with strongly unbalanced sample sizes at the stage of data collection has therefore a much greater potential to increase the accuracy of estimates than post-collection approaches. Gross (2002) suggested that this problem may be partly overcome by using previous knowledge of the relative importance of different life cycle transitions to sample individuals more efficiently. He also gives two examples showing that an alternative method, efficient sample composition (ESC), can considerably increase the precision of population growth rate estimates. However to apply the ESC-method, we need to make an educated guess of the relative importance of different transitions in the life cycle for population growth. This educated guess can be made using data from studies of the same or a related organism, the investigator's prior knowledge, pilot data or comparative demographic studies (Gross 2002), but we need to know how sensitive the sampling efficiency is to the appropriateness of the guess. Hence a method that is both simple to apply and does not require any previous knowledge of the demography, but that can provide significantly more precise and accurate measures than conventional methods would be of a large potential value.
In the first paper I explore this issue. Specifically I first examine how sensitive the ESC-method is to the quality of our previous knowledge of the species demography. Secondly, I investigated if a relatively simple approach, sampling an equal number of individuals per stage, can provide better estimates than plot-based sampling and if these estimates are as good as those based on the ESC-strategy.
To estimate the effectiveness of three different sampling strategies, the traditional plot based sampling, the ESC sampling and sampling equal number of individuals, on the accuracy and precision of properties of the resulting matrix, I used data for 32 different plant species collected from the literature.
The results show that the ESC method is sensitive to the reference matrix used to estimate the sampling proportions. In contrast, sampling equal numbers of individuals per stage provide estimates that were relatively robust and overall more accurate than both the ESC-method and the conventional plot-based method. I therefore conclude that collecting demographic data from an equal number of individuals per stage may constitute a simple and accurate method that is likely to improve the quality of demographic studies in many cases.
In the second paper I move on to studying local population dynamics. Specifically I compare population dynamics of two congeneric species that inhabit dry grasslands and differ in rarity, and I try to identify the key differences between them and thus to determine what factors are important in determining species rarity and commonness. Factors responsible for the commonness and rarity of closely related species have been a topic of many previous studies. However these studies suffer from two important drawbacks. (i) Most of the studies compare species that are closely related phylogeneticaly but occupy different habitats, so any observed difference may be due to habitat differences. (ii) Most studies also concentrate only on single life history traits, with unknown relevance for the population growth rates of the species, though knowledge of the growth rates is necessary to demonstrate that the factor is really responsible species rarity.
In the second paper I compare the complete demography of two Cirsium species sharing the same habitat, one of them being very rare and the other very common. Because seed herbivory is very common in this genus, I hypothesize that it may be one important factor in the rarity of the rare species.
The results show that population growth rate is slightly lower in the rare species, and this translates into a large difference in the local extinction probability. Seed predation does not differ between the species. However I demonstrate that in connection with the data on complete demography seed predation is the key factor causing the lower population growth rate in the rare species.
These results are the first estimation of factors responsible for commonness or rarity of plants in terms of population growth rate without confounding differences in habitat. It is also the first demonstration of differential effects of seed herbivory on the life cycles of two congeneric species.
Local population dynamics of species are also a topic of the third paper. Here I explore the effect of population size and genetic diversity on the performance of a plant species. Decreases in population size and genetic diversity are important concerns for conservation in fragmented landscapes. Therefore many studies have recently studied and demonstrated negative effect of low population size and genetic diversity on single plant traits. However nothing is known on the effects of these differences on the long-term probability of species survival.
In the third paper I study the effect of population size and genetic diversity on population growth rate in a rare perennial herb occurring in fragmented grasslands. Its performance was measured using several seed traits. The data were then connected with data on the mean demography of the species. Three different matrix models differing in the number of transitions based on measurements in a range of populations differing in size were used to explore the relationship between population size, genetic diversity and population growth rate.
The results show that genetic diversity itself is only weakly related to plant performance. It is however positively related to population size and population size is positively related to most of the seed traits. All the three matrix models showed that despite the decline in seed production in small populations, the population growth rate is always positive. The model using only data on seed production per flower head from each population in the matrix model, and the model using these data plus data on number of flower heads per plant and field germination rates indicated a positive relationship between population size and population growth rate. Contrary to this, according to model using realistic data on number of seeds per flower head and number of flower heads per plant in the matrix model no such relationship exists.
I conclude that declines in some plant traits in small populations do not necessarily lead to the species' direct endangerment. I thus suggest that wider use of matrix population models in this type of studies may provide new insight into the real effect of population size on species demography. The results also demonstrate that conclusions about the effect of habitat fragmentation on demography are crucially dependent on the parameters used to estimate plant performance and use of different parameters may lead to different conclusions.
Moving from the local scale to the regional scale incorporates questions about the importance of different local and regional processes for species distribution. This essentially means asking questions about what factors affect species distribution at different spatial scales. Answers to these questions enable determination of whether the distribution of a species is limited by availability of seeds and sites. Recently there have been an increasing number of studies concerned with the effect of various types of limitations on species' local population sizes and distribution patterns at the landscape scale. The terminology used to describe these limitations is however inconsistent. Since the terms are often used as a part of conclusions of papers, the inconsistency in their use obscures the message of these papers. In the fourth paper I thus review the current uses of these terms, identify the basic concepts involved in the discussion of a limitation and link the concepts with the term. Finally I discuss the experimental approaches that are used to assess these limitations.
I differentiated four basic concepts resulting from combinations of limitation by environment versus ability to grow and spread, and two spatial scales (local and regional). The two concepts at each spatial scale are expected to form a gradient of all possible combinations of the two respective types of limitations. In the considerations of various experimental approaches used to assess these limitations, I conclude that sowing experiments, meaning seed addition into existing population or seed introduction into unoccupied habitats, are the only reliable types of evidence for these limitations.
In the fifth paper I explicitly explore the above-mentioned limitations for species distribution. In concordance with paper four I start with the notion that distribution of species in fragmented landscapes is a result of combined seed and site availability at different spatial scales. At the local scale species may be limited either by seed numbers and germination ability (seed limitation) or by the availability of microsites suitable for germination (microsite limitation). At the regional scale species may be limited by ability to reach the site (dispersal limitation) or by availability of suitable habitat (habitat limitation).
While a lot is known on importance of these limitations for species distribution on one spatial scale, it is not known how their importance differs among different spatial scales. Additionally, while there is a lot of information on the effect of environmental factors on both pattern of seedling establishment and distribution of adult plants, the correspondence between these patterns is unclear, as is the effect of spatial scale on this relationship.
In this study I sowed seeds of seven species of dry grasslands into twenty-two localities that differed in occupancy by these species. I followed seedling establishment of these species over three years. The number of resulting seedlings was then compared at two regional scales, between occupied and unoccupied localities and occupied and unoccupied blocks within occupied localities, and at a local scale, between plots with and without seed addition within occupied blocks. Furthermore, I examined relationships among environmental factors and the number of seedlings and distributions of adult plants at the two regional scales.
The results show that both seed and microsite availability are important in structuring the distribution of these plant species. Their relative importance however depends on the spatial scale considered. Also the relationship between environmental factors and patterns of seedling recruitment and adult occurrence is clearly scale dependent.
Paper five shows that many species are dispersal limited and thus that there are suitable habitats that are currently unoccupied. This expectation is a prerequisite of studies of metapopulation dynamics of species. Using approaches such as that in paper five is the most direct approach to identify suitable unoccupied habitats. To get a good estimate of suitability using this method one should follow the whole life cycle of the species as population bottlenecks may occur at later stages of recruitment (Losos 1995, Gustafsson et al. 2002). This can, however, take much longer than any research project can last (Ehrlén and Eriksson 2000). Therefore, alternative indirect ways to estimate habitat suitability are sought (e.g. Husband and Barrett 1996).
It the sixth paper I therefore propose a new quantitative technique to identify suitable but unoccupied habitats for metapopulation studies in plants. It is based on species composition in a habitat and knowledge of species co-occurrence patterns. It uses data from a large phytosociological database as a background for estimating species co-occurrence patterns. If such a database is not available, the technique can still be applied using the data for which the prediction is done simultaneously to estimate the species co-occurrence pattern. Using the technique I was able to indicate suitable unoccupied habitats and differentiate them from the unoccupied unsuitable ones. I also identified habitats with low probability of being suitable that were occupied. Compared to a direct approach of identification of suitable habitats, which involves introduction of a species to the habitat and studying its performance, the approach presented here is much easier to apply and can provide extensive information on habitat suitability for a range of species with much less effort and time.
In the seventh paper I integrate the knowledge about local population demography of species (papers 2 and 3) with information on availability of suitable unoccupied habitats (papers 5 and 6) and with information on species dispersal ability and spatial structure of habitats in the landscape in a framework of a spatially explicit metapopulation model. I use this model to estimate the effect of habitat fragmentation on survival probability on a species occurring in fragmented grasslands. The motivation of this paper is that common approaches to estimating the effect of habitat destruction on species survival probability, such as metapopulation capacity, are based on the assumption that the current distribution of species in habitat fragments is a result of equilibrium between local colonization and extinction. This may be a reasonable assumption for short lived, highly dispersible organisms, but is unrealistic in sessile long-lived species such as most plants. Here I use an alternative approach, a realistic dynamic landscape-level model that does not use this assumption. It enables estimation of effect of habitat destruction using realistic field data on biology of a species and on landscape structure. Since the approach relies on direct comparison of changes in population size and survival probability due to habitat changes, it can be easily extended to other conservation questions, such as the effect of local population destruction or searches for optimal reintroduction strategies.
I apply this method to a perennial herb Succisa pratensis that is a typical representative of fragmented low-production grasslands. The results show that habitat destruction alone has only little effect on its prospects for survival. The effect however increases when combined with population destruction, which is expected to play a significant role in the study system. Using the same approach I was also able to design optimal reintroduction strategy for the species, assuming the species was extinct from the landscape. Given the biology of this species, I argue that only a dynamical model with local events as the one presented here makes it fully possible to evaluate its survival perspectives in fragmented landscapes and to design the most appropriate reintroduction strategies.

Conclusions

In this thesis I have investigated local and regional population dynamics of species of fragmented grassland habitats. I have explored both methodological and terminological aspects of this type of study (papers 1,4, and 6) and provided specific examples of factors affecting local (papers 2 and 3) and regional (papers 5 and 7) species dynamics.
From the methodological point of view this thesis has suggested a new way of sampling data for demographic studies (paper 1) that is expected to considerably increase precision of our knowledge on species population biology. It has also suggested a new method for identifying suitable unoccupied habitats for metapopulation studies (paper 6) that is expected to be much more easily applicable and at least as informative as the preceding ones. Difficulties in identifying suitable unoccupied habitats are one of the major concerns in application of metapopulation theory to plants (Freckleton and Watkinson 2002). Since this method provides an easy to apply tool for identification of suitable habitats, it has the potential to increase our understanding of metapopulation dynamics of plant species.
Another important part of this study is the exploration of local population biology of species in fragmented grasslands (papers 2 and 3). These papers used the methodology proposed in paper 1. Both of these papers explore important topics related to estimating of endangerment status of rare species. Both of the topics have been explored in many previous studies. The strength of the studies presented here compared to all the previous studies is the connection of studies on single species traits with species population models. By doing this I demonstrate the quantitative importance of changes in single traits due to low population size and genetic diversity for long-term population growth rate (paper 3) and to demonstrate which differences in single traits of species have the potential to result in differences in species commonness in the landscape (paper 2).
To connect the local population dynamics of species to their regional distribution I explored the importance of different factors determining species distribution on different spatial scales. Because the terminology used to denote these different types of limitations is very variable, I first identify the basic concepts in these studies, and linked them to appropriate terms and suggested methods of their study (paper 4). Then I use this approach and demonstrate the importance of different types of limitation for species distribution is scale dependent. This shows an important weakness of previous studies of the importance of different types of limitation for species distribution, since these are concerned with one spatial scale only.
In the final paper (paper 7) I perform population viability analysis of a perennial herb restricted to fragmented grassland habitats at a landscape level. The method applied here extends the traditional methodology of assessing the effects of habitat fragmentation on prospect for species survival that is based only on species distribution patterns and assumes that species distribution is in equilibrium. The major strength of the approach used here is that it is applicable also to non-equilibrium situations and it enables to fully use also information on local population dynamics. In this analysis I combine data on local species dynamics such as those gained in paper 2 and 3, with information on factors affecting species distribution at the landscape level analyzed in papers 5 and 6.
The results of this thesis suggest that both local and regional processes are crucial for dynamics of perennial plants in fragmented grassland habitats. They show that our knowledge of factors determining species dynamics in fragmented habitats is, in spite of the large number of studies on this topic, very incomplete. They also demonstrate that incorporation of knowledge on local population dynamics to studies of the importance of population size and genetic diversity for the prospects for species survival, on the factors determining species rarity as well as to studies on regional dynamics species can provide new valuable insight into these topics.