SEARCH

SEARCH BY CITATION

Keywords:

  • Ecological niche models;
  • fundamental niche;
  • realized niche;
  • scenopoetic variables;
  • species distribution models;
  • transference of distributions

Abstract

  1. Top of page
  2. Abstract
  3. SDM and ENM are not synonyms
  4. Fundamental niche
  5. References

A recent set of discussion papers in the Journal of Biogeography by McInerny and Etienne (henceforth M&E) questions the value of niche concepts in relation to a diverse group of practices collectively labelled species distribution modelling (SDM), and specifically the usefulness of the idea of a fundamental niche. In this Correspondence, I argue that certain types of SDM may indeed dispense with niche concepts, but that such is not the case for an important class of SDM-based activities, including transferring predictions in space and time. Using a single term (SDM) to denote diverse objectives and practices does not help to clarify issues; I discuss this point. I also review several criticisms raised by M&E about the use of the concept of fundamental niche in the context of modelling species' distributions and their environments.

Recently, McInerny and Etienne (‘M&E’) presented three reflection papers that I will call ‘Ditch’, ‘Stitch’ and ‘Pitch’ (McInerny & Etienne, 2012a,b,c) on the usefulness of the term ‘niche’ in ecology. They focus on the subdiscipline of species distribution modelling (SDM) in a broad sense, highlighting conceptual confusions surrounding use of the term ‘niche’ in this context, and aiming to provide a clarifying discussion. I believe that their papers will stimulate debate on what is indeed a conceptual mess. The three papers provide ample food for thought, and it is impossible to discuss them exhaustively in a brief comment. However, two points in particular caught my attention. First, throughout the papers, M&E fail to make explicit the important distinction between modelling the distributions of species and modelling the environmental conditions that allow species to persist, both referred to as SDM. The difference is important because M&E's arguments fit well to some activities denoted in ‘SDM’, but not at all to others. This point is clear when applying some of the ideas and schemes in Stitch to problems discussed in Ditch and Pitch. Second, in Ditch, M&E present an argument against use of Hutchinson's (1957) fundamental niche. This deserves an answer, since I believe that their criticism of what is a most useful theoretical concept for some activities in the field of ‘SDM’ is misguided. Hence here I comment on two issues mostly related to Ditch and Pitch, although I will refer to parts of Stitch.

SDM and ENM are not synonyms

  1. Top of page
  2. Abstract
  3. SDM and ENM are not synonyms
  4. Fundamental niche
  5. References

Just as ‘niche’ means different things to different people, ‘SDM’ also refers to a heterogeneous set of concepts and objectives. In a nutshell, the difference is illustrated using Figure 6a in Stitch, where the ‘abiotic environment’ is correlated with the distributional properties of a ‘species’. That is, the focus is on the species part of the diagram, and modelling of this should be called SDM. However, it is also possible to focus on the abiotic conditions part, and modelling this, being different, deserves a different term. The research objective may be on the distribution, or on the conditions, or on both. In other words ‘SDM’ methods and algorithms may be used to calculate: (1) maps of occupied distributional areas; (2) potential distributional areas; and (3) sets of environmental conditions (niches) corresponding to these areas (Peterson et al., 2011). However, throughout their contributions, M&E apply ‘SDM’ to any of the above, without making the distinction. For instance, SDM means environmental conditions on p. 2099 of Ditch and p. 2107 of Stitch, but refers to occupied areas in geographical space in Figure 1 of Ditch and on p. 2115 of Pitch.

Although usage of a single term to cover substantially different objectives has been defended on grounds of ‘neutrality’ (Elith & Leathwick, 2009), it can greatly hinder clarity. What is to be gained by referring to all the above with a single, ‘neutral’ term, when they are radically different (although related) concepts and objects? This is illustrated by the fact that it is possible to model the spatial pattern of a species' distribution without reference to environmental conditions (Bahn & McGill, 2007), and conversely, one can estimate favourable environmental conditions for a species without distributional data (Birch, 1953). Moreover, estimation (by correlative methods) of sets of environments as conditions within which species are found (‘look-up tables’, as M&E call them) is not synonymous with modelling actual distributions, as this goal would require adding data or hypotheses about at least dispersal (Barve et al., 2011), and ideally also effects of interacting biotic variables. When focus is on favourable conditions and their implications, ‘SDM’ deserves a different name: it has been called ecological niche modelling (ENM; Peterson, 2006), and correlative algorithms provide one avenue by which to estimate the realized scenopoetic niche (Austin et al., 1990), where the term scenopoetic (Hutchinson, 1978) refers to abiotic and non-interactive conditions, such as climate.

The consequence of failing to make these distinctions is illustrated well in Pitch (p. 2115). Towards the conclusion of their arguments, in discussing ‘the applied activity of species distribution modelling’ M&E state: ‘Despite some alliance with “niche” (Peterson et al., 2011), SDM is not dependent on “niche” (Elith & Leathwick, 2009).’ However, by the arguments above in perpetuating the practice of referring to a broad range of possibilities under the single term SDM, M&E hinder the important distinction that it is valid and quite interesting to estimate and understand objects that historically have been called niches (among other things). Moreover, people pursue such research without trying to find ‘ultimate definitions’ of niche. Several of M&E's criticisms and arguments indeed apply to correlational SDM in a strict sense (i.e. as a tool for modelling current geographical patterns in space), but not to SDM as process-oriented distribution modelling, and certainly not to ‘SDM’ as the study of the structure and dynamics of ranges of tolerance of species to environmental conditions (essential for transferring models in space or time). The last two require some notion of what is the range of tolerance to environmental conditions (niches), and these can be studied in the absence of competitors (fundamental niches). This is very useful when attempting a process-oriented modelling. Therefore, one can indeed ‘ditch the niche’ in some cases, but most certainly not in the most biologically interesting applications, i.e. those requiring an understanding of driving factors.

Fundamental niche

  1. Top of page
  2. Abstract
  3. SDM and ENM are not synonyms
  4. Fundamental niche
  5. References

In Ditch, M&E question the usefulness of the idea of Hutchinson's ‘fundamental niche’ in the context of species' distributions, mostly because of a perceived lack of connection with basic population dynamic processes. I will refer to these perceived shortcomings below, but first it should be stated that, despite being a very theoretical concept, the fundamental niche remains central for a comprehensive understanding of the factors affecting species' geographical distributions. The need for such a concept appears regularly in ‘SDM’ applications for invasive species, or in the context of climate change (Peterson et al., 2011), i.e. any time that instead of a simple geographical interpolation of points, one needs a transfer (in space and/or time) of a prediction. This is so because one can ‘transfer’ the species tolerances (i.e. the niche), but not the actual distribution. Although indeed highly abstract, fundamental niches can be partially studied under laboratory conditions (Birch, 1953), or derived from first principles of biophysics (Kearney & Porter, 2004) and projected spatially using a GIS (Kearney & Porter, 2004; Higgins et al., 2012), which means that they are far from being a purely theoretical notion.

M&E use Hutchinson's (1957) definition of the fundamental niche: a multivariate space ‘every point in which corresponds to a state in the environment which would permit the species … to exist indefinitely’, and mention that Hutchinson also established relationships between these multivariate spaces and geography (Colwell & Rangel, 2009) (an extremely fruitful idea), and then proceed with three arguments.

In their first argument against fundamental niches (Ditch, p. 2097), M&E state that such niches can be defined rigorously only for conditions, and not for resources ‘with which species interact’. This assertion is not strictly true, as demonstrated in detail by Chase & Leibold (2003), who show how to define fundamental niches based on interactive variables. However, the point is that most SDM/ENM is performed using precisely non-interactive variables and therefore, for the intended focus in their trilogy of papers (niche and modelling distributions), it would not matter if only non-interactive variables could define the fundamental niche because these are mainly the ones used. As they acknowledge explicitly, the scenopoetic variables of Hutchinson (1978) are not in an interactive, dynamic relationship with species’ populations, and these variables are not consumed, competed for, or modified by a species (i.e. they are conditions), which is why they adapt so easily to the disciplines of SDM/ENM (see Figure 6a of Stitch). Note that this does not mean in any sense that the details of population dynamics may not be affected by scenopoetic, non-interactive variables, as M&E discuss in their papers. For instance, anything affecting the growth rate may lead to chaotic behaviour for discrete-generation species. My point is simply that the fundamental niche can be defined using the variables most used in SDM/ENM. Perhaps in the future it would be possible to add interactive variables in most SDM/ENM studies and this may require either substantial modification of the fundamental niche concept, or even dismissing it entirely, but currently it is a useful idea, as long as one makes the distinction required by variable type.

The second argument (Ditch, p. 2098) is that fundamental niches need to be defined using variables that do not interact, in an ecological sense, with other variables, and gives the example of flowers as dimensions of the fundamental niche of pollinators. It is true that in the case of interactive bionomic niche variables (Hutchinson, 1978) dynamical coupling would complicate things significantly (see Chase & Leibold, 2003, for a treatment of this problem). Again, however, this point is irrelevant in most SDM/ENM applications, where, until very recently (see Stitch, p. 2108), interacting variables have not been used. Indeed, SDM/ENM can be based entirely on scenopoetic variables, with demonstrated – albeit obviously not universal – success (Peterson et al., 2011). This helped SDM/ENM to grow as disciplines, because data on scenopoetic variables are available in quantities measured in terabytes.

M&E's third argument against the concept of fundamental niches is based on whether complications of population dynamics are ignored in its definition. Of course, population dynamics determines the distribution (Soberón, 2010) and its complications would affect a general niche definition, as discussed (among many others) by Holt (2009), and in Stitch. However, at coarse resolutions, population dynamics is not necessarily relevant to SDM/ENM, for two reasons. First, a fundamental niche can be defined rigorously and operationally on the basis of how scenopoetic variables affect the intrinsic growth rate (Hutchinson, 1957; Holt & Gomulkiewicz, 1997; Pulliam, 2000), which is density independent. Simply stated, all combinations of environmental variables that permit a positive intrinsic growth rate define a fundamental niche; and its projection in geographical space, at a given time, represents a coarse-grained potential area of distribution for a species (Soberón, 2007; Colwell & Rangel, 2009; Peterson et al., 2011) regardless of population dynamics. Second, some major effects of population dynamics, such as dispersal limitations and metapopulation dynamics, can be dealt with in the context of reductions of the potential area corresponding to the hypothetical fundamental niche, using post-processing of simple correlative models (Saupe et al., 2012) or, recently, the hybrid models described in Stitch (p. 2108, and see Schurr et al., 2012).

Within regions where environments are favourable and intrinsic growth rates are positive, it is population ecology that provides the fine-grained structure of a distribution, but for the purpose of estimating distributions at geographical scales, the details of population dynamics over the entire range of a species simply may not be needed (Soberón, 2010). It is true that including interactive variables and increasing resolution remains a grand challenge for the discipline (Kissling et al., 2012; Wisz et al., 2013) (see Figure 6d and 6e in Stitch), and that tackling it correctly will be helped by making correct conceptual and mathematical distinctions, as outlined in Stitch, but most of SDM/ENM applications will still benefit significantly by the idea of a pre-interactive set of favourable environmental conditions, which constitute the fundamental niche.

M&E conclude their section on fundamental niches (Ditch, p. 2098) stating that ‘… fundamental niches are only coherent concepts when species are invariant to themselves, variables are invariant to everything else, and where direct mapping from fundamental to realized niche space is possible’. Well, the question then would be whether conceptual coherence under such apparently restrictive premises is not only achievable (which it is), but also theoretically interesting and useful in practice. It turns out that it is possible to characterize theoretically fundamental niches under those premises, and that this is rather helpful when interpreting the results of SDMs and ENMs, mainly when predictions are transferred in space or time (Peterson et al., 2011). Nevertheless, there is indeed a paradox in the wide use of SDMs and ENMs when their premises are apparently so restrictive. This paradox is what makes the need for specific and comprehensive theorizing about these premises so critical. I suggest that this theorizing may be better developed within the specificity of the problem (what determines an area of distribution, as a function of tolerances, movements and interactions), rather than from the, in my view, over-general perspective advocated by M&E.

Finally, studying fundamental niches requires ecophysiological thinking. This is central to a process-oriented, forward-looking approach to studying geographical distributions (Barve et al., 2011). This approach will continue to develop as an active topic of research because it provides biologically deep links between autoecology and biogeography, and opens doors to studying mechanistically the evolution of areas of distribution (Higgins et al., 2012). In the field of modelling species distributions, ditching the fundamental niche of Hutchinson would deprive researchers of a deep and powerful unifying concept. It is unlikely that this will happen.

To conclude, in Pitch, M&E review several books, noting that little use is made of the term ‘niche’. On the other hand, a Web of Science search for papers with the term ‘niche’ in their title (only in ecological journals) showed that, since the year 2000, their numbers have been growing exponentially at a rate of 0.17 per year (a doubling time of 4 years). Many ecologists are clearly not ditching niches. Nevertheless, M&E are correct that something indeed ought to be done about the conceptual mess surrounding the term. M&E's way is to equate ‘niche’ with ‘the ecology of a species’, a definition that ‘specifies no metrics, no parameters, and no particular models’ (Pitch, p. 2116), and to leave the messy details to models of specific instances of the general concept, as exemplified in Stitch. This possibility indeed is valid and logically coherent. Others, however, may prefer to constrain the scope of a theory to the specifics of the problem (the description, understanding and prediction of species' distributions as a function of niche, dispersal and interactions), and then develop it on the basis of available data, operational definitions and linking concepts using models. This type of restricted-scope, model-driven theory has been proposed in other areas of ecology (Martínez del Río, 2008). Whether this idea is an alternative to, or an instance of, what M&E recommend, I am not sure, but readers interested in niche definitions of less-than-universal scope can judge for themselves. Peterson et al. (2011) recently presented one such scope-delimited theory for niches and distributions, and for niches in community ecology one cannot do better than read Chase & Leibold (2003).

References

  1. Top of page
  2. Abstract
  3. SDM and ENM are not synonyms
  4. Fundamental niche
  5. References
  • Austin, M.P., Nicholls, A.O. & Margules, C.R. (1990) Measurement of the realized qualitative niche: environmental niches of five Eucalyptus species. Ecological Monographs, 60, 161177.
  • Bahn, V. & McGill, B.J. (2007) Can niche-based distribution models outperform spatial interpolation? Global Ecology and Biogeography, 16, 733742.
  • Barve, N., Barve, V., Jiménez-Valverde, A., Lira-Noriega, A., Maher, S.P., Peterson, A.T., Soberón, J. & Villalobos, F. (2011) The crucial role of the accessible area in ecological niche modeling and species distribution modeling. Ecological Modelling, 222, 18101819.
  • Birch, L.C. (1953) Experimental background to the study of the distribution and abundance of insects: I. The influence of temperature, moisture and food on the innate capacity for increase of three grain beetles. Ecology, 34, 698711.
  • Chase, J.M. & Leibold, M. (2003) Ecological niches: linking classical and contemporary approaches. University of Chicago Press, Chicago, IL.
  • Colwell, R.K. & Rangel, T.F. (2009) Hutchinson's duality: the once and future niche. Proceedings of the National Academy of Sciences USA, 106, 1964419650.
  • Elith, J. & Leathwick, J. (2009) Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics, 40, 677697.
  • Higgins, S.I., O'Hara, R.B., Bykova, O., Cramer, M.D., Chuine, I., Gerstner, E.-M., Hickler, T., Morin, X., Kearney, M.R., Midgley, G.F. & Scheiter, S. (2012) A physiological analogy of the niche for projecting the potential distribution of plants. Journal of Biogeography, 39, 21322145.
  • Holt, R.D. (2009) Bringing the Hutchinsonian niche into the 21st century: ecological and evolutionary perspectives. Proceedings of the National Academy of Sciences USA, 106, 1965919665.
  • Holt, R.D. & Gomulkiewicz, R. (1997) The evolution of species niches: a populations dynamic perspective. Case studies in mathematical modelling: ecology, physiology and cell biology (ed. by H.G. Othmer, F. Adler, M. Lewis and J. Dillon), pp. 2550. Prentice-Hall, Englewood Cliffs, NJ.
  • Hutchinson, G.E. (1957) Concluding remarks. Cold Spring Harbor Symposia on Quantitative Biology, 22, 415427.
  • Hutchinson, G.E. (1978) An introduction to population ecology. Yale University Press, New Haven, CT.
  • Kearney, M. & Porter, W.P. (2004) Mapping the fundamental niche: physiology, climate and the distribution of a nocturnal lizard. Ecology, 85, 31193131.
  • Kissling, W.D., Dormann, C.F., Groeneveld, J., Hickler, T., Kühn, I., McInerny, G.J., Montoya, J.M., Römermann, C., Schiffers, K., Schurr, F.M., Singer, A., Svenning, J.-C., Zimmermann, N.E. & O'Hara, R.B. (2012) Towards novel approaches to modelling biotic interactions in multispecies assemblages at large spatial extents. Journal of Biogeography, 39, 21632178.
  • Martínez del Río, C. (2008) Metabolic theory or metabolic models? Trends in Ecology and Evolution, 23, 256260.
  • McInerny, G.J. & Etienne, R.S. (2012a) Ditch the niche – is the niche a useful concept in ecology or species distribution modelling? Journal of Biogeography, 39, 20962102.
  • McInerny, G.J. & Etienne, R.S. (2012b) Stitch the niche – a practical philosophy and visual schematic for the niche concept. Journal of Biogeography, 39, 21032111.
  • McInerny, G.J. & Etienne, R.S. (2012c) Pitch the niche – taking responsibility for the concepts we use in ecology and species distribution modelling. Journal of Biogeography, 39, 21122118.
  • Peterson, A.T. (2006) Uses and requirements of ecological niche models and related distributional models. Biodiversity Informatics, 3, 5972.
  • Peterson, A.T., Soberón, J., Pearson, R.G., Anderson, R., Martínez-Meyer, E., Nakamura, M. & Araújo, M.B. (2011) Ecological niches and geographic distributions. Princeton University Press, Princeton, NJ.
  • Pulliam, R. (2000) On the relationship between niche and distribution. Ecology Letters, 3, 349361.
  • Saupe, E.E., Barve, V., Myers, C.E., Soberón, J., Barve, N., Hensz, C.M., Peterson, A.T., Owens, H.L. & Lira-Noriega, A. (2012) Variation in niche and distribution model performance: the need for a priori assessment of key causal factors. Ecological Modelling, 237–238, 1122.
  • Schurr, F.M., Pagel, J., Cabral, J.S., Groeneveld, J., Bykova, O., O'Hara, R.B., Hartig, F., Kissling, W.D., Linder, H.P., Midgley, G.F., Schröder, B., Singer, A. & Zimmermann, N.E. (2012) How to understand species' niches and range dynamics: a demographic research agenda for biogeography. Journal of Biogeography, 39, 21462162.
  • Soberón, J. (2007) Grinnellian and Eltonian niches and geographic distributions of species. Ecology Letters, 10, 11151123.
  • Soberón, J. (2010) Niche and area of distribution modeling: a population ecology perspective. Ecography, 33, 159167.
  • Wisz, M.S., Pottier, J., Kissling, W.D. et al. (2013) The role of biotic interactions in shaping distributions and realised assemblages of species: implications for species distribution modelling. Biological Reviews, 88, 1530.