Resolving the SLOSS dilemma for biodiversity conservation: a research agenda

Abstract

The legacy of the 'SL > SS principle', that a single or a few large habitat patches (SL) conserve more species than several small patches (SS), is evident in decisions to protect large patches while down-weighting small ones. However, empirical support for this principle is lacking, and most studies find either no difference or the opposite pattern (SS > SL). To resolve this dilemma, we propose a research agenda by asking, 'are there consistent, empirically demonstrated conditions leading to SL > SS?' We first review and summarize 'single large or several small' (SLOSS) theory and predictions. We found that most predictions of SL > SS assume that between-patch variation in extinction rate dominates the outcome of the extinction-colonization dynamic. This is predicted to occur when populations in separate patches are largely independent of each other due to low between-patch movements, and when species differ in minimum patch size requirements, leading to strong nestedness in species composition along the patch size gradient. However, even when between-patch variation in extinction rate dominates the outcome of the extinction-colonization dynamic, theory can predict SS > SL. This occurs if extinctions are caused by antagonistic species interactions or disturbances, leading to spreading-of-risk of landscape-scale extinction across SS. SS > SL is also predicted when variation in colonization dominates the outcome of the extinction-colonization dynamic, due to higher immigration rates for SS than SL, and larger species pools in proximity to SS than SL. Theory that considers change in species composition among patches also predicts SS > SL because of higher beta diversity across SS than SL. This results mainly from greater environmental heterogeneity in SS due to greater variation in micro-habitats within and across SS habitat patches ('across-habitat heterogeneity'), and/or more heterogeneous successional trajectories across SS than SL. Based on our review of the relevant theory, we develop the 'SLOSS cube hypothesis', where the combination of three variables - between-patch movement, the role of spreading-of-risk in landscape-scale population persistence, and across-habitat heterogeneity - predict the SLOSS outcome. We use the SLOSS cube hypothesis and existing SLOSS empirical evidence, to predict SL > SS only when all of the following are true: low between-patch movement, low importance of spreading-of-risk for landscape-scale population persistence, and low across-habitat heterogeneity. Testing this prediction will be challenging, as it will require many studies of species groups and regions where these conditions hold. Each such study would compare gamma diversity across multiple landscapes varying in number and sizes of patches. If the prediction is not generally supported across such tests, then the mechanisms leading to SL > SS are extremely rare in nature and the SL > SS principle should be abandoned.

Keywords: SLOSS cube hypothesis; dispersal; edge effect; extinction-colonization; geometric effect; habitat fragmentation; landscape scale; metacommunity; spatial sampling effect; species aggregation.

Fig 1

The total area contributing the species pool available to colonize a set of several small patches (A) is larger than the total area contributing the species pool available to colonize a set of few large patches of the same total area (B). Light‐coloured rectangles are patches. Boxes around them represent the areas from within which habitat can contribute colonists to the patches. Dark‐coloured rectangles are the areas of other habitat patches within the local landscape surrounding each patch.

Fig 2

When species distributions are clumped or spatially autocorrelated, a few large patches (A) will intersect (sample) fewer species than several small patches (B and C). Different colours represent different species within continuous habitat in a single ecoregion before habitat loss (large rectangles). Squares represent patches subsequently created by habitat loss. When the landscape extent (maximum distance between patch edges) is the same for few large and several small patches (A versus B), several small patches will cover the area more evenly and will therefore intersect more species: in A two large patches intersect three species while in B eight small patches intersect five species. This effect is accentuated if the several small patches are further apart than the few large patches: in C eight small patches intersect nine species compared to three species in A.

Fig 3

Illustration of the ‘SLOSS cube’, combining SLOSS‐relevant theory and empirical SLOSS studies. The axes are based on the theory and predictions summarized in Table 1. The proportional volumes of the three outcomes are based on their proportions found in a review of empirical SLOSS studies in which sampling effort was unbiased, i.e. sampling proportional to area (see fig. 2b in Fahrig, 2020): 50% SS > SL (yellow); 40% SL = SS (green); 10% SL > SS (blue). The SLOSS cube hypothesis predicts that SL > SS will dominate only when all of the following are true: between‐patch movement rate is low, the influence of spreading‐of‐risk on population dynamics is low, and across‐habitat heterogeneity is low, leading to low beta diversity.

Fig 4

SLOSS can be evaluated by comparing cumulative species richness across the same number of sample sites (black squares) randomly placed within habitat (green rectangles) in multiple landscapes of the same size, each containing the same total area of habitat, but distributed in different numbers and sizes of patches. Two example landscapes are shown here, each with 10 sample sites placed randomly in habitat. Note that when the landscape has many small patches, some will not be sampled. This is not a problem because the unit of analysis in such a study is the landscape, not the patch.