Does the Environment “Filter” or “Select” Species? Bridging the Ecologies of Microbes and Macro-Organisms for a Common Niche Assembly Theory

Abstract

More than five decades before the introduction of “environmental filtering” in plant and vegetation sciences, Baas Becking proposed that the “environment selects” for studies in microbial ecology. He coupled this with the ubiquity law that he proposed for microbes to obtain the tenet “everything is everywhere, but, the environment selects”. Nowadays, while this tenet is mostly used as a null hypothesis for studies of microbial biogeography, the latter part has large implications for niche assembly theories. In this respect, it is very similar to the idea of “environmental filtering”, although some minor differences exist regarding how both concepts have been applied in macrobial and microbial ecologies. During the second decade of the 21st century, the usefulness of the latter has been questioned due to difficulties in disentangling the roles of environmental (abiotic) and ecological interaction in community assembly. A new vision has emerged in the literature that considers the environmental filter as dynamic and continuously influenced by biotic communities. With a small modification, this scheme provides a solution that can accommodate the ecologies of both microbes and macro-organisms for a common niche assembly theory.

1. Introduction

That the environment plays a major role in shaping the species composition of communities in nature is largely accepted. Nevertheless, there has been a strong divide between the conceptual approaches developed in the domains of microbial ecology and of the ecology of macro-organisms, respectively [1]. The latter area is, hereafter, referred to as macrobial ecology. In this paper, I briefly review the different assumptions and approaches and I underscore that recent developments allow the divide to be bridged for a common framework.

In microbial ecology, the tenet “Everything is everywhere, but, the environment selects” was a powerful paradigm for many decades in the 20th century, when the study of the microbial ecology of prokaryotes was predominantly based on culturing studies. The basic ideas of this tenet were rooted in the research and teachings of Martinus Beijerinck (1851–1931), Professor of Microbiology in Delft (the Netherlands) from 1895 to 1921, although the actual quote was from Lourens Baas Becking (1895–1963) [2] and was originally published in 1934 in Dutch [3] as “alles is overal: maar het milieu selecteert” (see Figure 1). Before 2005, there was confusion about who the author was that formulated this tenet for the first time [2], probably because no full translation of the original work into English was available. Since then, the accessibility of Baas Becking’s main publications for the global scientific community has improved. First, his 1934 book was translated into English by Deborah Sherwood and Mishka Stuip and edited by Don E. Canfield, who also provided an extensive introduction, which was published in 2016 [4]. Second, in 2020, an improved and extended version of his book Geobiology, as an English manuscript, was discovered by Alexander Raat. During the Nazi occupation of the Netherlands, Lourens Baas Becking joined the resistance and undertook two attempts to escape to England, which both failed. As a result, he was imprisoned in 1944 in the Kriegs-wehrmachtgefängnis in Utrecht, where he worked on this manuscript with hardly any access to scientific literature. This latter manuscript was annotated and published in 2022 in Geochemical Perspectives [5] together with a biography of Baas Becking [6]. The first part of the statement “everything is everywhere” has also been referred to as the ubiquity law.

Being rooted in plant community studies and reinforced by trait-based approaches in macrobial ecology developed in the late 1970s, the concept of the environmental filter has appeared in plant and vegetation sciences since 1992 [7,8]. This corresponded to a desire to develop a predictive science of ecology with testable hypotheses. It has been recognized that the filter represents a metaphor, which can also be envisioned as a sieve. Hence, this approach was introduced for solving a problem, i.e., “It basically is a problem of deleting those species unsuited to a specified set of environmental conditions.” [7]. The environmental conditions characterizing the environmental filter have thus been restricted to abiotic variables. Once the question of eliminating the species that cannot live under the given abiotic conditions imposed by the environment was solved, it was thought that researchers could disentangle the impact of other drivers, such as dispersal and biotic interactions. This has resulted in adopting a conceptualization through a sequence of filters, with the environmental filter sandwiched between the top dispersal filter and the bottom interaction filter [9]. The latter refers to ecological interactions among organisms, which can be negative (competition) or positive for one (facilitation) or both (mutualism) of the interacting species, or beneficial for one and detrimental for the other (e.g., predation and parasitism).

The aim of the present study is to confront the visions of ‘the selection by the environment’ and the ‘environmental filtering’, as these are used in the microbial and macrobial ecologies, respectively. I describe the critiques in the literature with respect to these concepts and how a novel framework appeared that according my opinion can be used both by microbial and macrobial ecologists [10]. This novel framework can be further modified to accommodate for studies of succession and other phenomena, like multiple stable states, which are all based on assuming strong feedbacks of the biocoenosis on the environmental filter.

2. Differences and Similarities Between Selection and Filtering by the Environment

Apparently, the second part of the tenet of Baas Becking, i.e., “the environment selects”, and the “environmental filtering” of species, used separately by microbial and macrobial ecologists, are very similar. This is very nicely exemplified by the first phrase in the Abstract of [9]: “Environmental filtering, where the environment selects against certain species…”. Nevertheless, the tenet of “the environment selects”, according Baas Becking’s view, is based on the assumption that the organism is already present in the environment (ubiquity law), albeit often at vey low densities, and that it will only proliferate under appropriate environmental conditions. The environmental filter approach is more encompassing in that it considers both the organisms already present in the habitat and those that immigrate into the habitat from outside. Hence, the environmental filtering effect is based on negative net growth rates of the populations of species in that habitat, which result in the filtering out of these species [9]. The ubiquity law has never been an issue for macro-organisms, as dispersal limitation and biogeographic distribution patterns are generally well-accepted for macro-organisms.

In general, from its origins in the late 19th century until the 1970s, microbial ecology strongly relied on isolation and culturing experiments. This influenced attitudes that are particularly well expressed by the following statement from Baas Becking: “de veldoecologie kan ons geen enigzins complete milieubeschrijving leveren; dit voorrecht behoort aan het laboratorium” (“field ecology cannot provide us a rather complete description of the environment; this privilege pertains to the laboratory”) (Baas Becking 1934, see Figure 1, page 15, bottom lines [3]). This statement should be interpreted as meaning that only in the laboratory does the microbiologist have full mastery of the environmental conditions he imposes on his cultures to allow him precise study of the link between environmental variables and the growth of the microbial species. Therefore, elective cultures, as defined by Beijerinck, more commonly known as enrichment cultures, thus represent the key tool with respect to the tenet. Extreme environments (particularly methane-rich, hypersaline, and high-temperature environments) were considered as particularly interesting analogs of enrichment cultures [3]. It should be noted that in the 1930s, ecologists did neither possess large data bases nor the arsenal of multivariate statistics allowing them to infer links between environmental variables and the occurrence of species and their traits in complex communities in large databases. In contrast, the majority of studies using the environmental filter as a concept have been based on statistical inference.

3. Problems with the Baas Becking Tenet and the Environmental Filter

The development of the analysis of DNA sampled from the environment has revolutionized microbial ecology. Hence, it has become obvious that, so far, only a very small fraction of prokaryotes has become available in culture and could thus been validly described according the bacterial code of taxonomy. The majority of the procaryotes are, therefore, recognized as part of not yet cultured biodiversity [11] and in some cases only represented by a DNA sequence. The majority of the procaryotes are, therefore, recognized as part of not yet cultured biodiversity [11], and in some cases, are only represented by a DNA sequence. Beijerinck and Baas Becking were very optimistic that by carefully controlling the conditions imposed in enrichment cultures, it would be possible to promote growth and extract any novel species. In general, we are now aware that this is much more difficult than anticipated during the first part of the 20th century. A striking example was the difficulties in isolating square-shaped bacteria. Square bacteria were observed for the first time by Walsby in 1980 [12] in a hypersaline environment, and subsequent microscopic observations showed that they are widely distributed and common in these types of environments. However, the square bacteria were systematically overgrown by faster-growing species in classical enrichments of halophilic bacteria, such as those used by Baas Becking. Hence, it took 24 years for the first successful enrichment and isolation of a square bacterium, which was named Haloquadratum walsbyi. The specifics that enabled this achievement were that pyruvate was used as the substrate for enrichment and that agarose rather than agar had to be used for the plates [13]. This shows that Beijerinck and Baas Becking were overly optimistic regarding the technical possibilities of artificially creating environments that select for any species being searched for. Nowadays, culturing studies no longer occupy a dominant position in microbial ecology, which now also relies very heavily on metagenomics.

Based on the analysis of DNA sampled from the environment, in the early 2000s, several studies (cited in [1]) documented biogeographic patterns for some prokaryotes, which undermined the ubiquity law “everything is everywhere” [1]. Since then, many studies have addressed this question for many different species of prokaryotes and some small eukaryotes, while using the Baas Becking tenet to frame their studies. Some observations point to a near-ubiquitous distribution of studied species, while others explain patterns by dispersal limitation.

During the second decade of the 21st century, environmental filtering was criticized, particularly because of difficulties with inferring its role from observational data [8,9,10,14,15]. It is particularly difficult to disentangle the filtering effects of the environment from those of biotic interactions [8,9,10]. Furthermore, it has been reported that the outcome of competition between species can be influenced by environmental (abiotic) factors [8,9]. Hence, the abiotic factors of the environment strongly interact with the biotic interactions within communities. For a microbial ecologist, the distinction between abiotic and biotic factors is often meaningless, as illustrated by the following questions: In the bottom layer of a lake, should the high concentrations of hydrogen sulfide, produced by sulfate-reducing bacteria, be considered an abiotic or a biotic factor? In its top layer, are the low concentrations of dissolved inorganic nitrogen, which result from the uptake of these compounds by competing phytoplankton species, be considered abiotic or biotic? Strong interactions and feedback links from biocoenosis on environmental filtering have also been described for plant communities, i.e., realizing that mycorrhiza and microbes increase nutrient availabilities for many plants [16,17] and interactions among plants contribute to create local environmental conditions that may deviate from the incident abiotic conditions [17].

In 2017, a discussion on how to proceed with the filtering concept was initiated in the journal Trends in Ecology and Evolution (TREE). This was introduced by Cadotte and Tucker [9] with the thought-provocative title “Should environmental filtering be abandoned?”. Two papers [10,14] responded to this proposal, and Cadotte and Tucker [15] replied to the responses. The proposal by Thakur and Wright [10] conceptualizes a more dynamic process whereby the community in the habitat feeds back to the environmental filter to modify it and create novel opportunities for other species to pass the environmental filter and proliferate in the habitat. This modification is particularly the result of ecosystem engineers and niche constructors. I believe that this conceptual scheme also applies for micro-organisms and that it could be adopted by microbial ecologists, giving new momentum to Baas Becking’s tenet “the environment selects”.

4. Discussion

More than five decades before the concept of “environmental filtering” was introduced in the domain of plant and vegetation sciences, Baas Becking formulated this idea as “the environment selects”. In terms of micro-organisms, he connected this concept with the ubiquity law through the coordinating conjunction ‘but’ to obtain the tenet “Everything is everywhere, but, the environment selects” [2,3]. The effect of environmental filtering can also be considered with respect to the niche theory of Hutchinson [18], who differentiated between the fundamental niche and the realized niche of a species. The former describes a hypervolume for this species where it can potentially thrive and the latter describes where it actually occurs in nature considering that biotic processes, e.g., competitive exclusion, may limit its occurrence. Hence, the environmental filter will eliminate the species for which the environmental conditions are outside the fundamental niche. Baas Becking was also a precursor in this respect by defining a ‘potential milieu’ and a ‘natural milieu’ of the species [2,3], corresponding, respectively, to the fundamental and realized niches of Hutchinson. Nevertheless, Baas Becking was less aware of how ecological interactions could restrict the occurrence of species with respect to their fundamental niche and adopted the term ‘natural milieu’ as the habitats where the species actually occurs.

A long gap in time between appearances of the formulations of the “environmental filtering” and “the environment selects” is explained by a lack of exchanges between macrobial and microbial ecologists. Such exchanges have improved in recent times and currently, many projects consider micro- and macro-organisms together [14,16]. Nevertheless, despite the fact that life forms constitute a continuum, inherent and operational elements continue to create a divide [1] between micro- and macro-organisms (see

Table 1. Elements that explain the divide between microbial and macrobial ecology.

Microbes	Macro-Organisms	Consequence and Link to Community Assembly Theories
Reproduction mainly clonal with gene exchange and recombination, including lateral gene transfer	Mating and sexual reproduction (Ernst Mayr’s biological species concept)	Less relevant for niche assembly model; important for evolution and species concept
Taxonomy of prokaryotes is based on living isolated strains obtained in culture	Taxonomy based on type of specimens (e.g., plant herbarium)	Contribution of culturing studies of microbes challenged due to large fraction of not-yet-cultured biodiversity
Use of environmental DNA is mainly based on DNA extracted from living cells	Environmental DNA mainly relates to traces of organisms left in the environment, with whole genomes rarely being extracted	Microbial communities assembled from metagenomics (traits inferred from functional genes)
Ubiquitous according Beijerinck and Baas Becking	Biogeographic patterns	Baas Becking paradigm currently questioned—some evidence for biogeographic patterns among prokaryotes

). While some of the intrinsic differences between micro- and macro-organisms have a significant impact on the evolutionary processes of the species, I think that a dynamic vision of the environmental filter can be applied for both. It is a fact that the bacterial code requests pure cultures of living cells for species descriptions. This implies that only a small fraction of the species is properly described, which can pose a problem when considering an assembly of communities. As a result, microbial ecologists have adopted the study of environmental. DNA and can use metagenomics as a current tool for ecological studies, and culturing studies are less dominant nowadays. As mentioned above, Beijerinck and Baas Becking were overly optimistic about the potential of enrichment cultures and underestimated the problem of ecological interactions among species in these cultures. Finally, the observations showing that a number of microbes show biogeographic patterns that could be related to dispersal limitations invites us to consider metacommunities which can be coupled through exchange rates.

With a small adaptation, the model proposed by Thakur and Wright [10] can be adapted to handle both macro- and micro-organisms (see Figure 2). The essential point is that the environmental filter is dynamic and it is continuously influenced by the biotic communities (biocoenosis). At the start, which can be envisioned as a habitat without organisms, the environmental filter allows for occupancy in the habitat of small set of species from the regional metacommunity, for which the environmental (and still abiotic) conditions correspond to their fundamental niche, i.e., three species in the example of Figure 2. Among these three species’ ecological interactions allow two species to proliferate in the habitat (net positive growth rates), while one is excluded (net negative growth rates). This represents the “interaction filter” [9]. The proliferating species will constitute persistent populations in the habitat that will feed back to the environmental filter and modify it. This modification will be particularly strong when ecosystem engineers and niche constructors occur in the community. The modified filter will now allow other species to settle and establish.

In the community (Figure 2), it no longer makes sense to consider the environmental filter as purely abiotic. The process will dynamically continue, and the next step in such a sequence is described in Figure 3. I have particularly modified the scheme from Thakur and Wright [10] to indicate that species are recruited through the environmental filter only from the regional metacommunity. On a global scale, different metacommunities exist, and these can be coupled through exchanges. Obviously, exchange rates are often higher for smaller organisms, which may explain why some microorganisms are ubiquitous. I think that this model can be applied both for micro- and macro-organisms.

A most pertinent reply to the question of Cadotte and Tucker [9] “Should environmental filtering be abandoned?” would be “no, it remains useful, but you need to consider that the environmental filtering is strongly impacted by feedback from biotic processes”. The latter was clearly recognized in the reply by Cadotte and Tucker [15] as nonindependence of the environmental filter. Hence, this explains why efforts of statistical inference used to disentangle the environmental (abiotic) filter from the biotic filter are often a failure and should thus be abandoned. The links between the environmental filter and biotic processes are particularly intricate for mycorhiza [16] and the microbial species in microbiomes associated with macro-organisms [14,16]. Hence, in line with the concept of Figure 2, the presence of the host macro-organism contributes to modifying the environmental filtering, allowing the pertinent microbes to settle and establish in the community of the specific microbiome. Aguilar et al. [14] go further in highlighting that many of the macro-organisms obligately depend on a healthy microbiome to cope with the incident abiotic conditions, meaning that the pertinent micro-organisms reciprocally modify the environmental filtering for their host macro-organism.

While accepting that the feedback of the biocoenosis in the habitat can modify the environmental filter, changing the selection from unfavorable (filtering out) to favorable for certain species ([10] and Figure 2), it is obvious that the reverse can also occur. Both phenomena and their impact on the community are depicted in Figure 3. For example, in a growing young forest with a progressive closing of the canopy, the originally high-light-adapted plants in the undercover will be filtered out, while novel opportunities will be created for the shade-adapted plants. During this process, the microclimate also changes to deviate from the incident climate conditions. This is another example of how strongly the biocoenosis affects the local environmental conditions. With the additional adaptation, the conceptual scheme could be used to study succession, which is defined as a directional change in species populations, the community, and the ecosystem at a site following a disturbance [19]. In addition, this conceptual scheme is also coherent with the theory of multiple stable states [20], as this theory is based on the impacts of combinations of positive and negative feedback loops from the biocoenosis to the environmental conditions. Thus, while the negative feedback loops will tend to stabilize the communities, positive feedbacks may drive ecosystems towards regime shifts whereby the so far prevailing species will be filtered out and other species positively selected.

The approach described in this paper, which is based on adaptations of the proposal of Thakur and Wright [10], is rooted in the individualistic approach in plant ecology of Gleason [21] and clearly refutes the organismic approach of Clements [22], who considered vegetation as a superorganism. Nevertheless, it also clearly refutes the statement of Gleason that “a plant community is scarcely even a vegetational unit, but merely a coincidence’’ [21], cited in [17]. This is because the linkages between the biocoenosis and the environmental filter together with the ecological interactions among the individuals in the habitat create a system with emergent properties. For example, this allows us to clearly identify vegetation units and typologies of microbial communities. Hence, I think that this conceptual scheme represents a significant contribution to bridging the ecologies of microbes and macro-organisms for a common niche assembly theory.