One Hundred Years of Population Ecology: Successes, Failures and 7 8 8 9 the Road Ahead 9 10 10 11 11 12 12 13 Charles J

Integrative Zoology 2015; 10: 233–240 doi: 10.1111/1749-4877.12130

1 REVIEW 1 2 2 3 3 4 4 5 5 6 6 7 One hundred years of population ecology: Successes, failures and 7 8 8 9 the road ahead 9 10 10 11 11 12 12 13 Charles J. KREBS 13 14 14 15 Department of Zoology, University of British Columbia, Vancouver, Canada and Institute for Applied Ecology, 15 16 University of Canberra, Canberra, Australia 16 17 17 18 18 19 19 20 20 21 Abstract 21 22 Population ecology is the most mature of the three subdisciplines of ecology partly because it has a solid math- 22 23 ematical foundation and partly because it can address the primary questions of distribution and abundance with 23 24 experimental protocols. Yet there is much left to do to integrate our population knowledge into community and 24 25 ecosystem ecology to help address the global issues of food security and the conservation of biodiversity. Many 25 26 different approaches are now being developed to bring about this integration and much more research will be 26 27 necessary to decide which if any will be most useful in achieving our goals of explaining the changes we see in 27 28 the distribution and abundance of animals and plants. Food web ecology would appear to be the best approach 28 29 at present because it uses the detailed information of the population ecology of particular species in combina- 29 30 tion with data on consumer–resource interactions to apply to the applied problems of biodiversity conservation, 30 31 food security, pest management and disease prevention. If we can use our understanding of population ecology 31 32 to address the practical problems of our time in a creative way, we will benefit both the human population and 32 33 the Earth’s biodiversity. Much remains to be done. 33 34 34 35 Key words: abundance, geographic distributions, history of ecology, population models 35 36 36 37 37 38 38 39 INTRODUCTION 39 40 40 41 Population ecology has come a long way since the 41 42 days of Charles Elton and all the early ecologists (Crow- 42 43 croft 1991). Writing the history of how any science has 43 44 changed in a century is a difficult process because any 44 45 one individual sees only one slice of what has tran- 45 46 spired. (McIntosh 1985; Kingsland 2005). In this review 46 47 I will try to discuss the general problems that population 47 48 Correspondence: Charles J. Krebs, Department of Zoology, ecology addresses and how well it has accomplished 48 49 University of British Columbia, 6270 University Blvd., solving these problems over the past century. On the 49 50 Vancouver, B.C. V6T 1Z4 Canada. way I will point out what I think are the failures of the 50 51 Email: [email protected] past if only to try to avoid repeating them in the future. 51

1 The number of ecologists in the world continues to tique because the metric of “progress” is not quantified. 1 2 increase in an approximate exponential manner. Along Progress in any science depends in part on the num- 2 3 with this increase is a scientific fragmentation that must ber of scientists working in an area, the time scale of 3 4 occur because one person cannot know everything, so the events and processes that require explanation, and 4 5 we become aquatic ecologists or biological control spe- the money available to carry out the necessary experi- 5 6 cialists or ecological geneticists or conservation biolo- ments. It is relatively easy to point out that all 3 of these 6 7 gists, and we concentrate on one small part of the eco- components of the rate of progress are at the low end of 7 8 logical spectrum. This leaves open the problem of the scale for ecology. Funding for ecological research is 8 9 missing vital information that is buried in the literature growing but the problems facing us are perhaps growing 9 10 that now seems ancient, or in journals no longer pub- at an even faster rate. The bottom line is that if we are 10 11 lished or not available electronically, or in people work- convinced that ecology is achieving its goals too slow- 11 12 ing on the other side of the world and not publishing in ly, we require some suggestions for how to improve the 12 13 English. All of these issues make assessing the success- science. Peters (1991) offered some suggestions but re- 13 14 es and failures of ecology problematical. This apology ceived mainly brickbats in return (Lawton 1991). Eco- 14 15 now given, I will charge ahead to try to describe what I logical critics can be easily offended but the main point 15 16 think ecologists wish to achieve and how far they have of criticism is to generate discussion. 16 17 achieved it. I concentrate on population ecology, which 17 18 I know most about, but also because it is the best de- GENERALITY IN POPULATION 18 19 veloped of the ecological subdisciplines and significant- 19 20 ly ahead in understanding when compared with com- ECOLOGY 20 21 munity and ecosystem ecology. I will not say much Ecologists, like all scientists, wish to generalize their 21 22 about microbial ecology, a rapidly moving field about results, and this is highly desirable. The problem we 22 23 which I know little. face is that we do not know the domain of applicability 23 24 of any particular finding. While we have global theories 24 25 THE GOALS OF POPULATION that predominated in the 1960s and 1970s, we are now 25 26 26 ECOLOGY in a stage of global confusion which has arisen because 27 of detailed local studies that reach opposite conclusions 27 28 The goal of population ecology is to understand the about particular ecological processes. The classical ex- 28 29 29 reasons for the distribution and abundance of microbes, ample is the controversy over density-dependent regu- 30 30 plants and animals. This was the message of Charles El- lation of population size. The models are mostly sim- 31 31 ton (Elton 1927) and Andrewartha and Birch (Andre- ple but the empirical results are not and we do not know 32 32 wartha & Birch 1954). The best achievement would be what to do with populations that show density-vague 33 33 to be able to predict changes in distribution and abun- relationships (Strong 1986). Beautiful hypotheses and 34 34 dance as critical environmental factors change. How- ugly facts, Dennis Chitty labeled it (Chitty 1996). 35 35 What do we do with data that do not fit any particular 36 ever, prediction is not always possible, even if we have 36 hypothesis? There are two possible approaches. First, 37 a solid understanding of the mechanisms causing eco- 37 ignore the contrary data. In the words of one famous 38 logical changes in distribution and abundance. Simi- 38 ecologist to me years ago, “We do not refute our crit- 39 lar problems are well recognized in geology with regard 39 ics, we simply ignore them.” There may be several im- 40 to earthquake predictions (Uyeda 2013). The end goal 40 portant reasons for ignoring results (e.g. faulty equip- 41 of ecological research is both to protect the Earth’s bio- 41 ment, faulty experimental designs and faulty logic) and 42 diversity and to increase the quality of life for all people 42 these must be investigated. However, the second ap- 43 through ecosystem management. Both these goals can 43 proach is more useful in the long run: change the do- 44 be achieved only by managing people as well as under- 44 main of applicability of the hypothesis or reformulate it. 45 standing natural processes, goals that have long been 45 There is a good example of this approach in the evalua- 46 recognized in fisheries management (Hilborn 2007). 46 tion of the intermediate disturbance hypothesis for plant 47 There is a nearly universal critique of ecology which 47 populations (Kershaw & Mallik 2013) and whether this 48 48 suggests that our science is achieving these goals very hypothesis applies only in certain specific kinds of plant 49 49 slowly (Peters 1991; O’Connor 2000; Graham & Day- communities. 50 ton 2002). It is difficult to evaluate this general cri- 50 51 One key reason for contrary results is that the scale 51

1 of the study or the experiment is not properly designed ogy (Anderson et al. 2000; Ioannidis 2005), no one ser- 1 2 to test the question at hand. Very often we make these iously questions the value statistical methods have add- 2 3 design errors first and then only later discover the prob- ed to ecological research. 3 4 lems of interpreting the resulting data. This kind of 4 5 problem is clearly seen in studying the social organiz- Experimental approaches 5 6 ation of rodents in which data from small pens do not Population ecologists began to implement field ex- 6 7 mimic those obtained from large pens or open field periments in the 1950s, and the value of experiments 7 8 studies (Wolff & Sherman 2007). Landscape ecology has been recognized ever since. This revolution was 8 9 is difficult and we rarely have the resources to conduct closely tied to discussion in the scientific world about 9 these large-scale experiments (Lidicker 1995). Never- 10 how to make rapid progress in science and a general re- 10 theless, they must be done to reach generality in popu- 11 view of the best approaches to scientific inference (Platt 11 lation studies and methods are now being developed to 12 1964; Hilborn & Stearns 1982; Beck 1997). The key to 12 achieve this goal (Borer et al. 2014). 13 much of this discussion is the underlying need for rigor- 13 14 ous hypotheses with clear testable predictions that are 14 15 SUCCESSES IN POPULATION relevant to the question at hand, along with alternative 15 16 ECOLOGY hypotheses with non-overlapping predictions (Ander- 16 17 son et al. 2000). An important qualification is that “ex- 17 18 There are four broad categories of success that popu- perimental approaches” must not be confined to ma- 18 19 lation ecology has achieved during the past century and nipulative experiments, because the same approaches 19 20 I will review each of them in turn. can be used for natural experiments and for systems that 20 21 Mathematics of populations can only be observed, not manipulated (Diamond 1986). 21 22 The recognition that most ecological systems are open 22 23 Science progresses when it bonds with mathematics, systems has made the transfer of predictions made from 23 24 and population ecology is the strongest of the sub- closed systems difficult to evaluate (Oreskes 1998). For 24 25 disciplines of ecology in its mathematical development this reason many of the model system studies in popula- 25 26 (Ginzburg et al. 2007). The key to all of this is that there tion ecology were initially misleading. 26 27 is a mathematics of populations, developed largely by 27 human demographers followed by population ecolo- Applied ecology 28 28 gists. If we know the vital rates of any population we 29 From a practical point of view the achievements of 29 can calculate its growth trajectory. We do not need to population ecologists working in applied areas are stel- 30 talk in qualitative terms because we have quantitative 30 31 methods for population arithmetic (Caswell 2001; Ranta lar. Not only do they help to solve practical problems 31 32 et al. 2006). The importance of population mathematics of our society but they also provide an operational test 32 33 cannot be overemphasized because science is quanti- of ecological theory. Consider marine fisheries manage- 33 34 tative, and with quantitative methods you can find out ment. Nearly 100 years ago fishery ecologists began to 34 35 when your empirical results are quantitatively incorrect ask questions about optimal yield in marine fisheries 35 36 and then begin to ask why. (Graham 1935; Walters & Martell 2004). From this ear- 36 37 ly start when marine fisheries were underexploited to the 37 38 Statistical methods present day when many fisheries are fully exploited and 38 39 Closely aligned with population mathematics has some overexploited, a series of developments springing 39 40 been the development of statistical methods specific for from population arithmetic have aided management and 40 41 population ecology. Statistical estimation theory, typi- provided guidance for sustainable fisheries production 41 42 fied by Program MARK (Cooch & White 2010) and (Hilborn 2007; Hauser & Carvalho 2008; Costello et al. 42 43 Program DENSITY (Efford et al. 2009) has brought 2012; Hilborn 2012). 43 44 new rigor to asking specific questions about surviv- Wildlife management has been a second important 44 45 al and dispersal in marked populations. Statistical in- area in which population ecology and harvesting theory 45 46 ference theory has dramatically improved our ability to have worked together to produce sustainable hunting as 46 47 apply strong inference to population processes (Ander- well as improving freshwater fishing. Changes in wild- 47 48 son 2008), and Bayesian inference has provided new life populations caused by the elimination or reduction 48 49 approaches to data evaluation and estimation (Ellison of major predators have been implicated in a series of 49 50 1996; Hobbs & Hilborn 2006). While arguments have problems with overabundant wildlife (Letnic & Koch 50 51 raged about the utility and misuse of P-values in ecol- 2010; Estes et al. 2011; Newsome & Ripple 2014; Rip- 51

1 ple et al. 2014). The result for wildlife management is ogy for which clear empirical tests have not been speci- 1 2 that we know for the most part what population poli- fied. 2 3 cies ought to be implemented for sustainable manage- Figure 1 illustrates this problem with respect to the 3 4 ment, but in some situations political or social demands testing of model predictions. It does not matter for this 4 5 prevent their implementation. The same problem affects illustration whether the predictions come from a verbal 5 6 fisheries management. The ecology is sound but not al- model or a strict mathematical model. Figure 1a illus- 6 7 ways acceptable socially or politically. trates the general dilemma in any science when excep- 7 8 Conservation biology is a third critical area in which tions are noted to predictions. One has to choose wheth- 8 9 population ecology in cooperation with conservation er to abandon the hypothesis and develop another or 9 10 genetics has been able to produce practical recommen- whether to tinker with the hypothesis by restricting its 10 11 dations for protecting threatened and endangered spe- range of application and, thereby, saving the hypoth- 11 esis. Figure 1b gives a simple example from popula- 12 cies. Early accomplishments in conservation biology 12 tion growth theory. If the logistic equation does not de- 13 were successful because they focused on individual spe- 13 scribe the growth of a particular set of populations, we 14 cies and their interactions, and for many of the species 14 can modify the logistic model by restricting it to cer- 15 15 of concern there were detailed natural history and popu- tain classes of populations, or we can abandon the mod- 16 16 lation data available. More recent conservation work el and find a better one. Kingsland (2005) discusses this 17 has focused on communities and ecosystems, particu- logistic example in more detail and Caswell (2001) de- 17 18 larly in the era of climate change. Conservation biology scribes why matrix models are a better approach for this 18 19 operates with two well recognized paradigms defined by particular problem. 19 20 20 Graeme Caughley (Caughley 1994): the declining popu- The second general failure in population ecology has 21 lation paradigm and the small population paradigm. It been controversies over topics that lead nowhere. Re- 21 22 is an action-oriented subdiscipline and its recommen- ductionism versus holism is one philosophical example; 22 23 dations come from our accumulated ecological know- density-dependence is another example from ecology. 23 24 ledge. That conservation actions may not always suc- One possible solution that can help to stop these kinds 24 25 ceed is sometimes the result of ecologists diagnosing of wasteful controversies is to ask the question: What 25 26 the problem incorrectly but perhaps more often the fail- practical problems can be answered by this topic? Peters 26 27 ure of social pressures and political policies that are dif- (1991) suggests a similar question: What predictions 27 28 ficult to change. follow from this topic? Density-dependence has always 28 29 been a difficult issue, beloved of ecological modelers, 29 30 FAILURES IN POPULATION ECOLOGY distrusted by many field ecologists, an issue similar to 30 31 competition and niche overlap in community ecology. 31 32 We ecologists should not ignore the failures of the There is (in my opinion) little point in arguing about 32 33 past lest we repeat them, and I would like to summar- points that do not connect directly to measurable items 33 34 ize briefly the main problems that have affected popula- in the real world. Density-dependence in reproductive or 34 35 tion ecology particularly in the past 50 years. Four fac- mortality rates can be tested for in any population with 35 36 tors stand out. enough data, but such results do not lead anywhere until 36 the ecological mechanisms behind the relationships are 37 The first has been the failure to state clear hypoth- 37 understood. 38 eses and alternative hypotheses, along with the predic- 38 39 tions each entails. The excuse for this is typically that A third general failure in population ecology has been 39 40 we do not know enough to state clear hypotheses about the focus on short-term studies, typically demanded by 40 universities of PhD dissertations and by the government 41 the population under study. A simple example of this 41 research agencies who often assume that every prob- 42 is the hypothesis that “the abundance of population X 42 lem can be solved in a 3-year research grant. A correlate is limited by food supplies.” There are many observa- 43 of this issue has been a continued lack of appreciation 43 tions we can make that are consistent with this hypoth- 44 of the need for monitoring. This general failure is now 44 esis (e.g. “species X eats food Y which is relatively rare 45 rapidly disappearing, and the issue of climate change 45 in the habitat”). The key is to state the observations 46 has forced ecologists to think much more of long-term 46 that are not consistent with this hypothesis but are con- 47 studies and long-term monitoring (Lindenmayer & Lik- 47 sistent with one of the alternative hypotheses. This is 48 ens 2010; Lindenmayer et al. 2012). Short-term stud- 48 the method of strong inference presented by Chamber- 49 ies have the additional disadvantage of being confound- 49 lin (Chamberlin 1897) and emphasized by Platt (1964). 50 ed by transient events, and, thus, to reach conclusions 50 51 There are still too many hypotheses in population ecol- 51

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17 18 18 19 19 20 20 21 21 22 22 23 23 24 24 25 25 26 26 27 Figure 1 (a) A general diagram of the scientific method and possible ways of dealing with contrary results in any science. (b) One 27 28 example from population ecology to illustrate how ecologists reacted to discovering that the logistic equation population growth 28 29 model fits very few field data. Question marks indicate that one has to decide whether to continue to modify the logistic model (e.g. 29 30 the theta-logistic) (Clark et al. 2010) or abandon it for a more general model of population change such as matrix models (Caswell 30 31 2001). 31 32 32 33 33 34 34 35 35 that are unreliable and cannot be extrapolated in space fect is produced by the different factors under study. All 36 36 or time. The simple message is that more replication is the these failings can be brushed aside as minor details 37 37 needed. of little importance but in ecology more than ever we 38 38 need clear thinking and clear presentations of results. 39 A fourth common issue in ecology has been the fail- 39 40 ure to appreciate the assumptions behind statistical an- 40 41 alyses. In spite of the publication of many excellent KEY ISSUES FOR FUTURE RESEARCH 41 42 statistical texts we continually see tests of inference ap- 42 plied incorrectly because of the violation of assump- One of the main goals for population ecology in the 43 coming decade is to integrate its findings into commun- 43 44 tions. Many papers are full of statistical tests about hy- 44 potheses that every ecologist accepted long ago. Elegant ity and ecosystem ecology. I will argue that you can- 45 not conduct community and ecosystem studies without 45 46 AIC (Akaike information criterion) analysis tables are 46 sometimes presented to test trivial points, simultaneous- a firm foundation of the population ecology of the dom- 47 inant species. Many ecologists disagree with this state- 47 48 ly avoiding the difficult and important questions that do 48 need testing. Graphs continually appear in all our jour- ment and yet the past is not a good guide for success of 49 this contrary view. In the 1950s we thought we could re- 49 nals with no error bars, no indication of the units of 50 duce all of ecology to energy flow, so that species names 50 51 measurement, and no appreciation of how large an ef- 51

1 would become insignificant and knowing that they were timeframe of approximately 100 years. That we can- 1 2 herbivores or carnivores and how many calories they not dismiss these evo-eco issues is clear from our hu- 2 3 contained was enough to achieve ecological understand- man interactions with growing resistance to pathogen- 3 4 ing. The difficulty was that many communities and eco- ic bacteria and viruses (Andersson & Hughes 2011), 4 5 systems differ because of their evolutionary history, and similarly increasing resistance to rodenticides in rats 5 6 general models of community and ecosystem function (Buckle 2013) and the problems of herbicide-resist- 6 7 typically fail because they ignore the details of species ant weeds (Shaner 2014). Ecology and evolution are in- 7 8 interactions. Current attempts to use functional types are extricably tied together and we cannot use the excuse 8 9 falling into the same trap of assuming some commun- that this makes problems even more complex as a good 9 10 ality while ignoring evolutionary history (Ratnam et al. reason for ignoring evolutionary changes in ecologic- 10 11 2011; Boulangeat et al. 2012). These attempts are worth al interactions. However, the timeframe of evolutionary 11 2 12 continuing only if we can get beyond the range of R change in higher organisms may not mesh well with the 12 13 values > 0.5. timeframe of PhD dissertation and government research 13 14 There is no harm in trying to integrate community grants. We need to think of studies that will span the 14 15 ecology with population dynamics by ignoring the spe- 50 to 100 year timeframe both to capture possible evo- 15 16 cies’ peculiar traits, and this research is important to at- lutionary changes and to monitor long-term slow eco- 16 17 tempt. However, the most useful model that is now ap- logical shifts, a difficult issue that is rarely discussed. 17 18 parent is the food web approach (Memmott 2009; If the larger issues that face us as population ecolo- 18 19 Thompson et al. 2012). There are an array of relatively gists have been outlined, we must not forget there are 19 20 new methods now available for linking food webs with many components of what we do know that must be 20 21 the details of population ecology into the issues of com- studied further to discover the gaps that are still un- 21 22 munities and ecosystems. The key is to produce test- explored. While we recognize landscape ecology as 22 23 able food web models with clear mechanistic relation- critical for understanding population dynamics, it is 23 24 ships that flow from understanding population ecology even more critical for community and ecosystem dy- 24 25 (Thompson et al. 2012). New methods of network an- namics. Yet the resources needed for landscape ecology 25 26 alysis integrated with food web studies present oppor- are rarely available in spite of this need. If we are un- 26 27 tunities for advancement in our predictive understanding sure now of how much generality we can expect to find 27 28 of ecosystem and community changes (Olff et al. 2009). in population dynamics, we will need much further rep- 28 29 Many other key issues can be brought under the need lication of our experiments in many locations to be able 29 30 to integrate populations and communities. Invasive spe- to answer this question (Borer et al. 2014). In general, 30 31 cies are a major problem all over the Earth, and popula- we face all the same issues as other sciences in this re- 31 32 tion ecologists have had relatively poor success in pre- gard. While molecular biology is currently the queen 32 33 dicting which species are likely to be invasive in spite of the biological sciences, we recognize that even these 33 34 of much research (Jeschke et al. 2012). Reducing the scientists with all their resources have barely begun to 34 35 impacts of invasive species is another important focus put together the genetic patterns of life on Earth. We 35 36 of research that has been less successful than would be must continue to monitor, measure and experiment be- 36 37 desired, perhaps because of the need to put invasive fore we can achieve a synthesis of population, commun- 37 38 species into a clear community and ecosystem context ity and ecosystem ecology. 38 39 (Norbury et al. 2013). Disease is another critical focus, 39 40 and the spread of tropical diseases into temperate coun- ACKNOWLEDGMENTS 40 41 tries with climate change is a strong alarm call for more 41 42 interaction in understanding the nexus of interactions I thank my colleagues for their insights into these 42 43 that lead to disease transmission in plants and animals issues, but I respect their differences of opinion on 43 44 (Ostfeld et al. 2010; Xu et al. 2011). many of the questions raised here. In particular, I thank 44 45 Behind all of these issues is the larger long-term issue Judy Myers, Tony Sinclair, Rudy Boonstra, Stan Boutin, 45 46 of how much ecological interactions can change due to Zhibin Zhang and Jim Hone for sharing their thoughts 46 47 evolutionary changes. Do population ecologists need to about the future of ecological science. This research was 47 48 worry about microevolution while they juggle all these supported by the Natural Sciences and Engineering Re- 48 49 other factors we discussed previously? The simple an- search Council of Canada and the Institute for Applied 49 50 swer now is that we do not know how many ecologic- Ecology, University of Canberra. 50 51 al interactions are fostering genetic changes in the short 51

1 REFERENCES the world’s unassessed fisheries. Science 338, 517– 1 2 20. 2 Anderson DR (2008). Model Based Inference in the Life 3 Crowcroft P (1991). Elton’s Ecologists: A History of the 3 Sciences: A Primer on Evidence. Springer, New York. 4 Bureau of Animal Population. University of Chicago 4 5 Anderson DR, Burnham KP, Thompson WL (2000). Press, Chicago. 5 Null hypothesis testing: Problems, prevalence, and an 6 Diamond J (1986). Overview: Laboratory experiments, 6 7 alternative. Journal of Wildlife Management 64, 912– 7 23. field experiments, and natural experiments. In: Dia- 8 mond J, Case TJ, eds. Community Ecology. Harper 8 9 Andersson DI, Hughes D (2011). Persistence of antibi- and Row, New York, pp. 3–22. 9 10 otic resistance in bacterial populations. FEMS Micro- 10 Efford MG, Borchers DL, Byrom AE (2009). Densi- 11 biology Reviews 35, 901–11. 11 ty estimation by spatially explicit capture-recapture: 12 Andrewartha HG, Birch LC (1954). The Distribution 12 likelihood-based methods. In: Thomson DL, Cooch 13 and Abundance of Animals. University of Chicago 13 EG, Conroy MJ, eds. Modeling Demographic Pro- 14 Press, Chicago. 14 cesses in Marked Populations. Springer, New York, 15 15 Beck MW (1997). Inference and generality in ecology: pp. 255–69. 16 Current problems and an experimental solution. Oi- 16 17 kos 78, 265–73. Ellison AM (1996). An introduction to Bayesian infer- 17 ence for ecological research and environmental deci- 18 Borer ET, Harpole WS, Adler PB et al. (2014). Finding 18 sion-making. Ecological Applications 6, 1036–46. 19 generality in ecology: A model for globally distribut- 19 20 ed experiments. Methods in Ecology and Evolution 5, Elton C (1927). Animal Ecology. Sidgwick and Jackson, 20 21 65–73. London. 21 22 Boulangeat I, Philippe P, Abdulhak S et al. (2012). Im- Estes JA, Terborgh J, Brashares JS et al. (2011). Trophic 22 23 proving plant functional groups for dynamic models downgrading of Planet Earth. Science 333, 301–6. 23 24 of biodiversity: At the crossroads between functional Ginzburg LR, Jensen CXJ, Yule JV (2007). Aiming the 24 25 and community ecology. Global Change Biology 18, “unreasonable effectiveness of mathematics” at eco- 25 26 3464–75. logical theory. Ecological Modelling 207, 356–62. 26 27 Buckle A (2013). Anticoagulant resistance in the United Graham M (1935). Modern theory of exploiting a fish- 27 28 Kingdom and a new guideline for the management of ery, and application to North Sea trawling. Journal 28 29 resistant infestations of Norway rats (Rattus norvegi- du Conseil Permanente International pour l’Explora- 29 30 cus Berk.). Pest Management Science 69, 334–41. tion de la Mer 10, 264–74. 30 31 31 Caswell H (2001). Matrix Population Models: Con- Graham MH, Dayton PK (2002). On the evolution of 32 32 struction, Analysis, and Interpretation, 2nd edn. Si- ecological ideas: Paradigms and scientific progress. 33 33 nauer Associates, Sunderland, MA. Ecology 83, 1481–9. 34 34 Caughley G (1994). Directions in conservation biology. Hauser L, Carvalho GR (2008). Paradigm shifts in ma- 35 Journal of Animal Ecology 63, 215–44. 35 36 rine fisheries genetics: Ugly hypotheses slain by 36 Chamberlin TC (1897). The method of multiple work- 37 beautiful facts. Fish & Fisheries 9, 333–42. 37 ing hypotheses. Journal of Geology 5, 837–48 (re- 38 Hilborn R (2007). Defining success in fisheries and con- 38 printed in Science 148: 754–59 in 1965). 39 flicts in objectives.Marine Policy 31, 153–8. 39 Chitty D (1996). Do Lemmings Commit Suicide? Beau- 40 Hilborn R (2012). The evolution of quantitative marine 40 tiful Hypotheses and Ugly Facts. Oxford University 41 fisheries management 1985–2010. Natural Resource 41 Press, New York. 42 Modeling 25, 122–44. 42 Clark F, Brook BW, Delean S, Reşit Akçakaya H, Brad- 43 Hilborn R, Stearns SC (1982). On inference in ecology 43 shaw CJA (2010). The theta-logistic is unreliable for 44 and evolutionary biology: The problem of multiple 44 modelling most census data. Methods in Ecology and 45 causes. Acta Biotheoretica 31, 145–64. 45 Evolution 1, 253–62. 46 Hobbs NT, Hilborn R (2006). Alternatives to statistical 46 Cooch E, White GC (2010). Program MARK: A Gen- 47 hypothesis testing in ecology: A guide to self teach- 47 tle Introduction, 9th edn. http://www.phidot.org/soft- 48 ing. Ecological Applications 16, 5–19. 48 49 ware/mark/docs/book/. 49 Costello C, Ovando D, Hilborn R, Gaines SD, De- Ioannidis JPA (2005). Why most published research 50 findings are false. PLoS Medicine 2, e124. 50 51 schenes O, Lester SE (2012). Status and solutions for 51

1 Jeschke J, Gómez Aparicio L, Haider S et al. (2012). Olff H, Alonso D, Berg MP et al. (2009). Parallel eco- 1 2 Support for major hypotheses in invasion biology is logical networks in ecosystems. Philosophical Trans- 2 3 uneven and declining. NeoBiota 14, 1–20. actions of the Royal Society B 364, 1755–79. 3 4 Kershaw HM, Mallik AU (2013). Predicting plant di- Oreskes N (1998). Evaluation (not validation) of quan- 4 5 versity response to disturbance: Applicability of the titative models. Environmental Health Perspectives 5 6 intermediate disturbance hypothesis and mass ratio Supplements 106, 1453–60. 6 7 7 hypothesis. Critical Reviews in Plant Sciences 32, Ostfeld RS, Keesing F, Eviner VT, eds (2010). Infec- 8 8 383–95. tious Disease Ecology: Effects of Ecosystems on Dis- 9 9 Kingsland SE (2005). The Evolution of American Ecolo- ease and of Disease on Ecosystems. Princeton Uni- 10 10 gy, 1890-2000. Johns Hopkins University Press, Bal- versity Press, Princeton, New Jersey. 11 timore. 11 12 Peters RH (1991). A Critique for Ecology. Cambridge 12 Lawton J (1991). Predictable plots. Nature 354, 444. 13 University Press, Cambridge, UK. 13 14 Letnic M, Koch F (2010). Are dingoes a trophic regula- Platt JR (1964). Strong inference. Science 146, 347–53. 14 15 tor in arid Australia? A comparison of mammal com- Ranta E, Lundberg P, Kaitala V (2006). Ecology of Pop- 15 munities on either side of the dingo fence. Austral 16 ulations. Cambridge University Press, Cambridge, 16 Ecology 35, 167–75. 17 MA. 17 Lidicker WZ Jr (1995). The landscape concept: Some- 18 Ratnam J, Bond WJ, Fensham RJ et al. (2011). When is 18 thing old, something new. In: Lidicker WZ Jr, ed. 19 a ‘forest’ a savanna, and why does it matter? Global 19 Landscape Approaches in Mammalian Ecology and 20 Ecology and Biogeography 20, 653–60. 20 21 Conservation. University of Minnesota Press, Minne- 21 Ripple WJ, Beschta RL, Fortin JK, Robbins CT (2014). 22 apolis, pp. 3–19. 22 Trophic cascades from wolves to grizzly bears in Yel- 23 Lindenmayer DB, Likens GE (2010). Effective Ecologi- 23 lowstone. Journal of Animal Ecology 83, 223–33. 24 cal Monitoring CSIRO Press. Collingwood, Victoria. 24 Shaner DL (2014). Lessons learned from the history of 25 Lindenmayer DB, Likens GE, Andersen A et al. (2012). 25 herbicide resistance. Weed Science 62, 427–31. 26 Value of long-term ecological studies. Austral Ecolo- 26 27 gy 37, 745–57. Strong DR (1986). Density-vague population change. 27 28 McIntosh RP (1985). The Background of Ecology: Con- Trends in Ecology and Evolution 1, 39–42. 28 29 cept and Theory. Cambridge University Press, Cam- Thompson RM, Brose U, Dunne JA et al. (2012). Food 29 30 bridge. webs: Reconciling the structure and function of bio- 30 31 31 Memmott J (2009). Food webs: A ladder for picking diversity. Trends in Ecology & Evolution 27, 689–97. 32 32 strawberries or a practical tool for practical prob- Uyeda S (2013). On earthquake prediction in Japan. 33 33 lems? Philosophical Transactions of the Royal Soci- Proceedings of the Japan Academy, Series B 89, 391– 34 34 ety B 364, 1693–9. 400. 35 35 Walters CJ, Martell SJD (2004). Fisheries Ecology and 36 Newsome TM, Ripple WJ (2014). A continental scale 36 Management. Princeton University Press, Princeton, 37 trophic cascade from wolves through coyotes to fox- 37 New Jersey. 38 es. Journal of Animal Ecology 84, 49–59. 38 39 Norbury G, Byrom A, Pech R et al. (2013). Invasive Wolff JO, Sherman PW, eds (2007). Rodent Societies: 39 40 mammals and habitat modification interact to gener- An Ecological and Evolutionary Perspective. Univer- 40 41 ate unforeseen outcomes for indigenous fauna. Eco- sity of Chicago Press Chicago. 41 42 logical Applications 23, 1707–21. Xu L, Liu Q, Stige LC et al. (2011). Nonlinear effect of 42 43 O’Connor RJ (2000). Why ecology lags behind biology. climate on plague during the third pandemic in Chi- 43 44 The Scientist 14, 35. na. PNAS 108, 10214–9. 44 45 45 46 46 47 Cite this article as: 47 48 48 49 Krebs C (2015). One hundred years of population ecology: Successes, failures and the road ahead. Integrative Zo- 49 50 ology 10, 233–40. 50 51 51