001 Trends in Ecology and Evolution

The Unified Neutral Theory of Biodiversity (UNTB), proposed as an alternative to niche theory, has been viewed as a theory that species coexist without niche differences, without fitness differences, or with equal probability of success. Support is claimed when models lacking species differences predict highly aggregated metrics, such as species abundance distributions (SADs) or species area distributions (SARs). Here, I summarize why UNTB generates confusion, and is not actually relevant to niche theory (i.e. an explanation for why and how many species coexist). Equal probability is not a theory, but lack of one; it does not include or exclude any process relevant to coexistence of competitors. Models lacking explicit species can make useful predictions, but this does not support neutral theory. I provide s suggestions that could help reduce confusion generated by the debate. The UNTB is not a null model for niche differences Proponents of the Unified Neutral Theory of Biodiversity (UNTB [1] describe its history as 'controversial'. Despite this recognition, there has not been much real discussion of the issues by proponents. Advocates do not dwell on literature that rejects the UNTB, focusing instead on UNTB predictions of species abundance distributions (SADs or other highly aggregated summaries. Here, I provide a perspective on this history as one who is a critic, but not to balance the perceived successes of proponents with a review of failures. I think more progress can come from highlighting where the confusion lies, what perpetuates it and steps that can be taken to move beyond it. The continuing confusion comes from misidentifying what is controversial and the fundamental coherence problem with UNTB. Rather than review the UNTB history I use some recent examples to highlight how confusion can perpetuate. The debate is not as hard to resolve as the literature suggests. There are simple steps that can be taken to steer the debate in a productive direction, but it has to start with proponents showing that they can respond to empirical and theoretical problems raised repeatedly in a large literature. I provide several suggestions following a summary of coherence and how it is being sidestepped with debate focusing on issues that are not controversial. The coherence problem The 'coherence problem' is a term I will use for the issues raised in [2] and [3]. This is not the same problem that has led others to reject UNTB; it is not about the fact that predictions are incorrect ([4-8]), not diagnostic of anything ([9-11]) or simply implausible The coherence problem is this: contrary to claims, neutral models do not assume that species occupy the same niche, have the same fitness or are the same in any sense; UNTB has nothing to say about how species differences affect diversity. Niche and neutral models differ only in terms of how the random elements in models are interpreted When species are at the same relative abundance, assigning identical parameter values to species in stochastic models allocates to each species on average 1/K of the total K niches or sites. When species have different abundances, the probability is weighted by relative abundance, because birth rates are assumed to scale with abundance. Drift is slow, not because species are the same, but rather because the average number of 'niches', 'sites' or 'success' opportunities for each species is the same, changing only as they change in abundance. Rather than make species the same, neutral models provide maximum success opportunities to each species. The fraction of sites a species wins depends on specific assumptions of a given model (e.g. a birth-death process or lottery). However, with respect to species differences, such models have nothing to say about the underlying cause (Box 1). Models that generate 'niches' at random or assign species to 'sites' at random do not represent different processes. They are not two alternative models, one 'niches-for-all', the other 'niches-for-none'. The subjective probability interpretation of a neutral model ('each species could win with probability p = 1/K 0 ) is not a statement of equivalence; it describes the state of maximum uncertainty. (For an event with K potential outcomes, complete ignorance is p = 1/K.