<h3>Excerpt</h3> <div><div> <h2>CHAPTER 1</h2> <p><b>Classifying Rodent Population Changes</b></p> <br> <p><i>Key Points:</i></p> <p>• Rodent populations vary from relatively stable to highly cyclic.</p> <p>• Populations of the same species may fluctuate strongly in one part of their geographic range and minimally in other areas.</p> <p>• Critical methodological issues regarding size of the sampling area, the sampling interval and length of study, the methods of estimation, and the habitat mosaic of the study site affect the value of any particular study.</p> <p>• The time scale of rodent dynamics is monthly or even weekly, and annual census estimates are of limited use for deciphering population processes.</p> <p>• The key question for all rodent population changes, whether cyclic or relatively stable, is what factors determine the population growth rate.</p> <br> <p>Small rodents have been a focus of interest for humans throughout history because of the diseases they have carried and their impact as pests of agricultural crops. The rise and fall of rodent populations have become part of myth and legend, from the Pied Piper and the Black Death to lemming migrations and rodents falling from the sky during storms. Underlying each of these legends is some germ of truth, and science in general seeks to uncover these germs and explain how they came to be. Population fluctuations or "cycles" became noticed when these natural history observations were analyzed by early ecologists. When Robert Collett in Norway and Charles Elton in Britain began to write about periodic eruptions of lemming numbers in Norway and vole numbers in Britain, few could imagine the complexity their insights would later uncover (Collett 1895; Elton 1924).</p> <p>Population ecologists have carried out more research on small rodent dynamics than on the dynamics of most other ecological groups. This is partly because rodents are convenient for graduate theses and other short research projects, and partly because of the damage rodents inflict on forest regeneration (e.g., Huitu et al. 2009) and agriculture (e.g., Singleton et al. 2005). Reviews of rodent population changes have become more specialized in light of the increasing literature, and therefore I think it critical to step back and take a broad overview of the current state of progress and the uncertainties and controversies that continue in this area of research. Progress has been uneven in understanding rodent population changes, partly because of bandwagons that develop and partly because of geography. Rodents in many parts of the world are poorly studied while those in other areas, like the North Temperate Zone of Europe and North America, have provided the bulk of our information base.</p> <p>In this chapter I will explore the various population changes that have been described for small rodents, and attempt to classify them into four categories of population patterns. There are two approaches one can take to any set of ecological patterns. First, one can assume that all patterns in the set are variants of a single general pattern, and then search for a single general explanation. In principle, this is what ecologists have done with population regulation. All populations are regulated, this approach asserts, and the regulation is produced by some type of density dependence in the broad sense. In principle, if the density dependence has some type of delay built in, the populations affected will fluctuate periodically, and may become "cyclic." But if the density dependence operates with little time delay, the population may show numerical stability or only seasonal fluctuations. In this simple yet powerful model, all types of population traces can be envisaged, and the ecologist's problem becomes one of how to find the density-dependent mechanism behind the population events (Krebs 2002). Does birth rate change? Does juvenile survival change? Do more individuals emigrate as density rises? These kinds of questions become the critical focus of ecological investigations.</p> <p>An alternative approach is to view each of the uncovered patterns as distinct, and then to develop a separate theory for each recognized pattern. For example, stable populations can be viewed as one particular case of rodent population dynamics, and explanations for stability would become the focus of these studies. Alternatively, cyclic populations can be viewed as a second pattern, and a different explanation would be proposed to cover that case. Clearly, these two different approaches to population analysis may converge, but the philosophical differences that underlie them will have a major effect on our ability to develop general theory and general models. One particular example of the alternative approach was developed by Turchin et al. (2000) in an analysis of vole and lemming cycles in Fennoscandia (box 1.1).</p> <br> <p><b>Background Issues</b></p> <p>Four background issues must be reviewed briefly before we can begin to classify patterns of rodent population changes. Throughout this book we will talk about populations and population density, and discuss hypotheses about regulation.</p> <br> <p><i>What Is a Population?</i></p> <p>We assume that we can define a population of small rodents unambiguously; this assumption is common in population ecology. In practice we define the population operationally by the area we choose to study: often a small trapping area of less than a few hectares. We implicitly assume that the population we are studying is closed—or, if it is open, that it has equal amounts of emigration and immigration across its boundaries (White et al. 1982). We also assume that the population we are studying is typical, and although statisticians caution us to use random sampling, virtually no one selects their study population site randomly. We assume that a site is representative of a broad class of local populations in the region with absolutely no idea whether or not this is correct.</p> <p>Not all rodent populations are closed, and if we have an open population we will find it difficult to measure survival rates, because of movements in and out of the area. In principle we can solve the problem of open populations by adopting a larger study site, but one soon runs into practical problems of time and budget. Consequently, we pray that our population is closed or nearly so. Few ecologists seem to lose any sleep over this fundamental issue.</p> <p>Most studies take little note of the landscape surrounding the study area; this is an issue that has recently come under scrutiny (Delattre et al. 009). We tend to assume that the population under study is little affected by events in the surrounding landscape—an assumption that conservation ecologists never make (Ferraz, et al. 2007), and which we will explore later in chapter 6.</p> <br> <p><i>What Is Population Density?</i></p> <p>Once we get over the first hurdle of defining a population, we run into a second problem of measuring its density. This involves two issues. First we need to measure population size accurately, and then we need to know exactly the area on which this population lives, so that we can measure abundance per unit area. This measurement is simple only with island populations, when we can sample an entire island.</p> <p>There are now numerous statistical techniques that can be used to estimate absolute population size (Krebs 1999; Pollock et al. 1990), and there is an extensive literature on this particularly for small mammals (e.g., Parmenter et al. 2003; Efford 2004). There is no excuse for rodent ecologists not to use these methods. Nevertheless, many studies continue to be published without adequate statistical methods of estimation, and much of the population data we will use in this book can be described only as indices of abundance, with all the problems associated with the use of indices (Anderson 00 ). If there is one practical recommendation for future work on small rodents, it is simply to use modern methods of population estimation.</p> <p>Once one has good estimates of population size from Program MARK (White 008), there is still the problem of estimating population density. The two alternative methods of density estimation from mark-recapture data are boundary strip estimation to determine effective grid size (e.g., Jett and Nichols 1987) and spatially explicit methods developed by Murray Efford (Efford 2004; Borchers and Efford 2008; Efford et al. 2009). Efford's suggestion was to fit a simple spatial model of animal trapping that estimated the probability of an animal being caught in a trap at a given distance from its home range center. It requires data on recaptures of individually marked animals to estimate the spatial model, but appears to be the best method now available for density estimation in live trapping studies of small rodents (Krebs et al. 2011).</p> <p>Given that we have good population estimates and a good measure of the area used by the rodents, a third problem arises: What part of the population is being sampled? We do not usually count babies in the nest as part of the population; and juveniles must reach a minimum size before they can be trapped, so they are also underrepresented in population measures. The age structure and, in particular, the number of breeding adults should be used when possible, so that density estimates have some biological reality. We assume in most studies that we are discussing the adult segment of the population and most of the younger age classes are not included. Rarely is this stated explicitly.</p> <br> <p><i>What Is the Time Step of Sampling Population Attributes?</i></p> <p>The biologically relevant time step over which to measure population attributes is on the order of a few days or weeks. But when we look for long-term data on population changes, we find that one data point per year, or two data points, represents much of these data for rodents. There are almost no data on winter populations for species that live in cold climates, due to the obvious difficulty of gathering such information. Nevertheless, these limitations on available data are often forgotten. The assumption that, for example, winter losses occur at a constant rate over the snow period in northern voles and lemmings may be convenient, but it must be recognized as a very large assumption.</p> <p>The advantages of using small rodents as experimental animals in field studies of population dynamics are offset by the requirement to sample at a relevant time frequency. It should be a basic principle of ecological experimental design that the sampling frequency reflects the timing of biological events rather than human convenience. For small rodents this frequency is on the order of two to three weeks, which for many species is the gestation interval and slightly less than the generation time.</p> <br> <p><i>What Data are Needed to Test Hypotheses about Population Limitation?</i></p> <p>When we proceed to test a particular hypothesis about rodent population changes, we adopt a uniformity-of-nature assumption: that whatever variable we can find to explain population changes will be the critical variable at other times and in other places. This assumption of repeatability is rarely tested, and is common to all of science but nevertheless worth remembering.</p> <p>Whatever hypothesis or model of rodent population change anyone wishes to test, it is critical that they have a set of explicit predictions for the hypothesis. There are excellent sources for discussions of experimental design and hypothesis testing in science, and thus no need to belabor the subject here (Anderson et al. 2000; Eberhardt 2003). The history of studies of rodent population changes can be read as a haltingly slow progression of making more precise predictions from hypotheses about limitation and regulation. One of the goals of this book is to try to make these predictions more explicit, and to review which predictions seem to fit the available data and which are still untested.</p> <p>The use of mathematical models to explore hypotheses of regulation in vertebrates has had the beneficial effect of making hidden assumptions more explicit (Berryman 2002; Turchin and Batzli 2001). This is a most desirable development, particularly because it can start a feedback cycle between empirical investigations and mathematical models, to the benefit of both.</p> <br> <p><b>Empirical Patterns of Population Change</b></p> <p>It is useful to look at some good examples of long-term data on population abundance for several rodent species before we begin to discuss the possible ways of classifying patterns of change. I have picked out four data sets to illustrate the patterns we need to consider. All of these data sets will reappear in later chapters as we discuss mechanisms of population change.</p> <br> <p><i>Lemmings in Siberia and Norway</i></p> <p>Russian ecologists have provided some of the longest time series of fluctuations of small rodents. The relative abundance of the collared lemming <i>(Dicrostonyx torquatus)</i> and Siberian lemming <i>(Lemmus sibiricus)</i> for 41 years is shown in figure 1.1 from Kokorev and Kuksov (2002). Fourteen peak years were described in 41 years of study, and almost all the fluctuations are periodic with a three-year interval. These lemming data may be considered typical of the classical three- to four-year vole and lemming cycle that Elton (1942) described many years ago. The pattern is clear but the quantitative dynamics are missing because these are indices of abundance, and we cannot interpret an increase from two to three in the index as being a numerical rise equivalent to a change from three to four.</p> <p>The Norwegian lemming <i>(Lemmus lemmus)</i> has been monitored in the mountains of south central Norway for more than 30 years (Framstad et al. 1993, 1997, personal communication). Figure 1. shows that these lemming populations in the mountains of southern Norway have fluctuated periodically, with three- to four-year intervals during this period. Again, the lemming population estimates are an index, but in this case snap traps were used to provide a catch-per-100-trap-nights index. We do not know if snap-trap indices show a linear relationship to absolute densities, although ecologists studying small rodents often make this assumption. Under that assumption, lemming cycles at Finse vary considerably in amplitude, so that cyclic amplitude is not a constant. These lemming data illustrate one common pattern of population change: cyclic fluctuations with a three- to four-year period. Not all lemming populations show this pattern (Reid et al. 1997), but many do.</p> <br> <p><i>Gray-Sided Voles on Hokkaido, Japan, and at Kilpisjärvi, Finland</i></p> <p>Gray-sided voles (<i>Myodes</i> [<i>Clethrionomys</i>] <i>rufocanus</i>) are the most common voles in the coniferous forests of Hokkaido, Japan. Because they eat the seedlings of forest trees in plantations, the Forestry Agency on Hokkaido began in 1954 to conduct censuses on 1,000 areas all over the island. Until 1992 these censuses were carefully carried out, proving to be a gold mine of population data on this species for 1 years (Saitoh et al. 1998; Bjørnstad et al. 1999). Population abundance was measured by snap trapping, and these data are thus an index rather than an estimate of absolute density. Figure 1. gives two of the 225 time series that have been analyzed in detail by Takashi Saitoh and his colleagues (e.g., Saitoh et al. 1997, 1998). These two locations are 200 km apart, and both populations fluctuate but do not always reach a peak in the same year (e.g., peaks reached in 1981 and not again until 1983). Populations around Heian appear to fluctuate more strongly than those at Ebishima. Peak populations occur at three- to five-year intervals, and these peaks are not as easily defined as those observed in lemming populations. Amplitudes of fluctuation are highly variable, as we found for Norwegian lemmings.</p> <p>The gray-sided vole occurs across Asia and Fennoscandia, and by chance it has also been the subject of another long-term study in Finnish Lapland at Kilpisjärvi. Kalela (1957) began this work, and it has been continued by several Finnish ecologists (Henttonen 1986 and personal communication). Population abundance has been measured by kill trapping with the results shown in figure 1.4. Peak populations occurred at intervals of three to five years, and during the 1990s there was a decrease in the amplitude of fluctuations. A similar pattern was found by Hornfeldt et al. (2005) in northern Sweden, who suggested that cycles had been dampening out in northern Fennoscandia during the previous 20 years. This issue was explored further by Ims et al. (2008), and we will discuss it further in chapter 6. High amplitude fluctuations returned in 2010, so the decrease in amplitude was only temporary (Brommer et al. 2010).</p> <p>Both <i>Microtus pennsylvanicus</i> and <i>Microtus ochrogaster</i> occur in grasslands of the central United States, and Lowell Getz and his colleagues studied populations of these two microtines from 1972 to 1997 (Getz et al. 2001). Their study sites were on farmland in central Illinois, and three habitats were monitored: tallgrass prairie (the original habitat of this region), bluegrass (<i>Poa praetensis</i>, an introduced grass), and alfalfa (<i>Medicago sativa</i>). Each of these habitats occupied about four to five hectares in a farming area dominated by corn and soybean crops. Live mark-and-release trapping was carried out monthly throughout the 25 years of study, so that absolute density estimates were available for a detailed picture of seasonal and yearly dynamics. These data represent the most detailed long-term data on vole populations in North America. </div></div><br/> <i>(Continues...)</i> <!-- Copyright Notice --> <div><blockquote><hr noshade size="1"><font size="-2">Excerpted from <b>Population Fluctuations in Rodents</b> by <b>CHARLES J. KREBS</b>. Copyright © 2013 by The University of Chicago. Excerpted by permission of The University of Chicago Press.<br/>All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.<br/>Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.</font><hr noshade size="1"></blockquote></div>