Peter Greig-Smith

Greig-Smith’s first published paper dealt with the varying growth forms of the dune grass Ammophila arenaria in calcareous sand dunes on the Isle of Harris, Outer Hebrides.[3] This paper was followed two years later by a detailed description of the dune’s plants,[4] using quantitative methods to describe the plant community. These early papers already mark his interest for quantification of natural phenomena. In this morphometric analysis of Ammophila arenaria, for example, Greig-Smith measured in detail the internode length of the plant’s rhizomes and noted the species’ tendency to form tussock-like clumps in older dunes and longer, exploratory rhizomes in unstable, younger dunes.

Some of Greig-Smith’s early work was devoted to taxonomy and biogeography, trying to bring a quantitative approach to these branches of the biological sciences. Much of his taxonomic work was devoted to the evolutionary taxonomy and ecology of pounceworts (Lejeunaceae, a family of liverworts).[5][6][7] In 1950 he published a synthesis paper on the taxonomy and biogeography of liverworts,[8] as an early attempt to shed some light on the question of the effects of Quaternary glaciations on the history of the British flora.[9] He tried to evaluate how many of the extant British liverworts had survived Pleistocene glaciations in situ and how many, in contrast, were postglacial migrants from southern latitudes. Almost two decades before the theory of Pleistocene refugia was proposed by Jürgen Haffer for the Amazon,[10] Greig-Smith was using plant distributions to understand the vegetation changes induced by Pleistocene glaciations, and setting a new direction for Quaternary Paleobotany that few could value in its full significance at that time. As part of the Flora of the British Isles series published in the Journal of Ecology, Greig-Smith drafted a synthesis paper of the natural history of the nettles of the United Kingdom, and a detailed analysis of the communities where they prosper.[11]

In 1940, Paul Richards, Sir Arthur Tansley, and Alexander Watt published a seminal paper[12] on the classification of tropical rainforests revealing a fascination with the complexity of tropical vegetation that was halted by the war, but that Watt passed down later to Greig-Smith, his student at Cambridge. In July 1948 Greig-Smith received a research grant from the Colonial Office to visit the island of Trinidad to study tropical rainforests. Encouraged by E. Ashby, an early pioneer of quantitative methods who was then professor of botany at Manchester, he left the university for six months to study secondary succession in Trinidad, in collaboration with colleagues of the Imperial College of Tropical Agriculture on the island.

Like Darwin in the Brazilian Atlantic Forest, he was gripped by the complexity of tropical forests. In order to try to unravel and explain this complexity, he developed novel statistical methods to test hypotheses on the association between species and the spatial pattern of saplings and adult trees in the field. An avid naturalist, he also developed in Trinidad a passion for orchids, which led him later to start cultivating his own orchid collection in the United Kingdom. After his visit to Trinidad in 1948, Greig-Smith drafted two detailed papers on secondary tropical forests.[13][14] The first study included a number of techniques that at that time were novel in vegetation analysis, such as the species-area plots developed by Ronald A. Fisher and Henry Gleason, size-specific survivorship, and quantitative floristic tables. In the second Trinidad paper, he also used nested analyses of variance (ANOVAs), comparisons with the Poisson distribution, and variance-to-mean ratios to test and understand spatial patterns in the trees of the secondary forest. He concluded his paper questioning Frederic Clement’s concept of plant communities as complex supra-organisms, and in support of Gleason’s individualistic model of plant communities.

In the early 1950s, Greig-Smith started using graphs in which he plotted the variance in plant density versus the size of the sample plot, to find peaks that could identify clump size in aggregated distributions. In doing so, he realized that ecosystem properties are scale-dependent: a species can have a regular distribution at a given scale, but be highly clumped at another.

Greig-Smith was fascinated by the phenomenon of plant spatial pattern, defined by the question: why are some plants regularly spaced while others are densely clumped? He realized that plants could be clumped at a certain scale, but that those clumps, in turn, could be randomly, or even regularly distributed at larger scales. Patterns and processes in ecology, he concluded, can have different behaviors at different scale. Strongly influenced by his early experience in agricultural experiments using random blocks and nested designs, he sought to answer this question by using, as his core tool, a series of quadrats nested within each other at different scales. Much of Greig-Smith’s published work was devoted to advancing the theory and methods of plant pattern, using his nested-quadrat approach as a basic tool. His fundamental idea that the perceived properties of ecological phenomena are scale-dependent set the foundations for the development of a theory of scaling in ecology, a central concept in modern climate change and global sustainability studies.

One of Greig-Smith’s early research papers, Gemmell, et al. (1953),[16] dealt with the formation of tussocks in Ammophila arenaria and included detailed observations on the clumped distribution of tillers, which form tussock-like clumps in older, fixed dunes but are regularly distributed in the younger, moving dunes. After this initial study, the spatial pattern of dunes and grasslands became a lifelong interest in his ecological research, published in a number of papers with collaborators and students.[17][18][19][20][21][22][23] One of Greig-Smith’s students, Andrew Morton, continued Greig-Smith’s pattern analysis studies in the dunes at Newborough Warren in Anglesey, Wales, near Bangor.

Below only the study of coastal dunes, Greig-Smith’s interests were strongly focused on tropical ecosystems. After his visit to Trinidad as a young researcher, he developed an interest for African savannas, on the possibility of using his methodology for pattern analysis to understand the importance of intraspecific competition in tropical, strongly rainfall-seasonal ecosystems.[24][25]

In different papers,[26][27][28][29] he showed the importance of scaling in the measurement of plant spatial pattern. He understood that researchers sampling the same vegetation with quadrats of different size could reach diametrically opposed conclusions: a researcher using small quadrats would find that his quadrats harbor either clumps or no plants at all, and would conclude that the species pattern in space follows a clumped distribution. Similarly, a researcher using large quadrats would find a similar number of clumps in each quadrat and would conclude that the species distribution is markedly regular. In reality, both would be right because a species can have a strongly clumped distribution at a certain scale and a regular distribution at another. As president of the British Ecological Society, he wrote an address in 1979,[30] in which he summarized his ideas on the importance of scaling in ecological spatial pattern. Some of his students took his ideas and pursued them further: Mark Hill developed a rigorous mathematical theory for the analysis of pattern at a community level,[31] while Kenneth Kershaw, after publishing in co-authorship with Greig-Smith, continued exploring for years the drivers of pattern at different scales.

The fundamental concept that ecological phenomena could have different behaviors when measured at different scales became really important for understanding large-scale phenomena in the biosphere. The Theory of Scaling became a mainstream concept in ecological research in the late 1980s and 1990s, with the publication of two now-classic papers: Wiens 1989[32] and Simon Levin 1992.[33] Greig-Smith, however, had figured out the main concepts four decades before, and both Wiens and Levin recognized him as a scientific pioneer.

Greig-Smith was among the first ecologists to understand that multivariate methods were destined to become important tools in quantitative plant ecology. His inclusion of multivariate classification and ordination in his book opened the door for the explosion of analytical approaches and methodological discussions that surged in the 1970s and 1980s. As with pattern analysis, Greig-Smith was ahead of his time: although the potential of multivariate methods—based on matrix algebra—was already recognized in the 1950s, mostly as a result of a growing need for more-rigorous methods in ecology, performing an analysis of a modestly sized floristic matrix of, say, fifty sites and one hundred species was a complicated and enormously time-consuming task. The generalized access to electronic computers, which began in the 1960s, opened the doors for the widespread use of these methods; Bangor’s first computer became available in 1964. Greig-Smith saw in multivariate methods an opportunity to analyze and understand the complex floristic composition of tropical forests, which had fascinated him since his visit to Trinidad in 1948. In 1963 he applied for a grant to investigate the use of quantitative methods to establish whether the highly diverse tropical rainforests showed any organized pattern, particularly in relation to environment, an open question at the time. With these funds, in 1964 he appointed Michael Austin as a postdoctoral research assistant to the project. Three seminal papers were published as a result of their collaboration, highlighting the potential of multivariate methods to analyze floristic patterns in complex, multispecies communities.[34][35][36] The power of his approach, merging multivariate descriptive methods with experimental hypothesis testing, is especially noticeable in the work of some of his students like Mike Austin (cited previously), Exequiel Ezcurra,[37] and F.B. Goldsmith,[38][39] among others. Mike Austin later moved to Australia and became one of the country’s leading ecologists.

Having laid the foundations of the discipline of multivariate methods, as a committed naturalist and field biologist Greig-Smith never took much interest in abstruse theoretical discussions, always expressing his concern about the risk of becoming too engaged with theoretical refinements of methodology and repeatedly stating his belief that numerical methods are worth developing only if they are to be used on real data in attempts to answer real questions with relevance in the field. The ultimate evidence, for him, was not in the computer output but in the field. In a synthesis paper published in 1980, he insisted that multivariate methods are simply hypothesis-generating procedures, and counseled students to do experimental tests addressing the insights from the classification and ordination analyses rather than taking the results of their analyses as an unchallengeable truth.[40]

Greig-Smith’s later papers on multivariate methods, with Mike Swaine[41] and Carlos Montaña,[42] were devoted to exploring new and innovative applications for multivariate analysis, including the analysis of time-series data and the exploration of the niche preference of species within a community. In the first two decades of the 21st Century, with the advent of powerful computers and large datasets from molecular genetics studies, the methods of numerical classification and eigenvector ordination, promoted by Greig-Smith half a century earlier, exploded in their use to become indispensable tools in the new molecular approaches to taxonomy, genomics, phylogeography, and anthropology.

Greig-Smith’s most influential work was his book Quantitative Plant Ecology, used by generations of students and young researchers around the world. The book, first published in 1957, enjoyed two subsequent editions in 1964 and 1982.[43][44][45]

The success of Greig-Smith’s book triggered the publication of many other similar books, many of them by former Greig-Smith collaborators (such as Kershaw 1964[46] and Orlóci 1975[47]), and nurtured an explosion of concepts and ideas around the quantitative treatment of ecological data, many of which arose in Canada (such as Pielou 1969[48]) and Australia (such as Williams 1976[49]).

In the 1950s and 1960s, at the time when the School of Plant Biology at Bangor consolidated as a global magnet for plant ecologists, an active scientific debate, excellently synthesized in Barbour (1996),[50] was taking place worldwide on the role of holism and reductionism as the main tools to understand ecological phenomena. In Britain, this debate centered mostly on the use of descriptive methods or the quest to understand patterns in nature, as opposed to the use of experimental methods or the search for ecological processes. In Bangor, in particular, the most vocal proponent of an experimental and reductionist approach to ecological research was John L. Harper, who often criticized quantitative vegetation scientists on their descriptive, rather than experimental, approach to research. Roy Turkington[51] recalls that Harper strongly favored the power of simple, direct experiments and distrusted complex multivariate methods, often arguing that ordination methods were “a formalized description of consequences.” “Nowhere is this holism more apparent than in the way ecology is taught in many schools and universities with its emphasis on complex systems and the ways in which they can be described,” wrote Harper in “After Description”,[52] and he immediately added: “There is probably no other science in which students are taught by being dropped into the deep end of complexity.” The contrast between experimental ecology in simplified systems, with only one or a few species, against large-scale pattern-seeking methods led Peter Greig-Smith and John Harper into frequent discussions in the form of good-humored debates following departmental seminars in Bangor, which, as amusingly evoked in Hill (2003)[53] and Matlack (2009),[54] were conducted amid legendary clouds of pipe smoke for the benefit of the students, who could see firsthand how academic arguments conducted both with rigor and humor can be very stimulating and deeply entertaining.

Greig-Smith, however, was not really a holist. From his very first papers on the tropical forests of Trinidad, he embraced Henry Gleason’s “individualistic” model of plant communities. Tim Allen, a Greig-Smith student in the 1960s, recalls that while in the 1980s “the use of descriptive multivariate methods degenerated into a ritual that community ecologists were expected to perform”,[55] Greig-Smith insisted that ecological research should always have experimental tests addressing the insights gained and the hypotheses derived from the quantitative analysis, as in Morton (1974).[56] Perhaps the main difference with Harper’s single-species experimental approach lied in Greig-Smith’s interest in complex, multispecies systems, and his opinion that projects should start with observations and the search for patterns, before establishing hypotheses about the processes that drive ecosystem functions and experimenting to tests these hypotheses. The debates between Harper and Greig-Smith at Bangor had a large influence on the development of ecological theory in the last four decades of the 20th Century.

In Bangor, Greig-Smith took the view that it was up to his students and young postdocs to define the project, develop the scientific hypotheses, and do the work, provided that they included quantitative ecology, his own area of expertise, in their graduate research. He rarely put his own name on his students’ research papers, so that his enormous influence is visible often only in the acknowledgements of published papers. However, some of the papers written by students and researchers while at Greig-Smith’s lab rank among the highest cited in ecology, and owe much of their originality to Greig-Smith’s mentorship.[57][58][59][60][61][62][63][64]