If something seems too good to be true…

Sure, small patches are important, but that important?

This post is about patch size and species diversity (yes, again), but I want to start with an analogy.

Vit Brunner CC BY-NC-SA 2.0

Imagine you drive to work and you have a choice of two different roads to get there. They are the same total distance and they have the same number of intersections, each with a set of traffic lights. However, one of the roads is a major arterial road, and the traffic lights default to green. The other is a bit of a back street and the lights are usually red. You probably wouldn’t be that surprised that the first route gets you there quicker? Well, I’m going to argue that ecologists (including me) have been doing something like this since the 1980s. I’ll circle back to this – let’s talk some ecology.

Cue diagram.

Figure 1. SLOSS analysis for three datasets, showing the possible outcomes (A) Large-to-small lies always above small-to-large implying greater importance of large patches, (B) overlapping curves suggesting patch size makes no difference, (C) Small-to-large lies always above large-to-small implying small patches are of greater importance.

Let’s say we have species lists for a group of habitat patches that differ in size. Say, all the bird species found on every island of an archipelago. We are not particularly interested in how many species we find on a single island, but in how many species we see in groups of islands for their total area. Especially when we combine groups of smaller, versus groups of larger islands. Essentially this is a question about the effects of habitat sub-division. One way it has been approached is by combining the species found on the islands in order of their size (Fig. 1), from the smallest to the largest (small-to-large) and from the largest to the smallest (large-to-small). Then we can plot the number of species we obtain as the islands are combined against the total area those islands collectively contain (hereafter SLOSS analysis).

I have a history with this particular pattern – in fact, my desire to understand them explains a lot about the direction of my PhD research. I now think I can explain them pretty well. I also think they are a classic example of how our eyes can deceive us.

It is a simple premise: if you add up the number of species in a group of small islands (or more generally, habitat patches) and there are more than you find in one or more large patches for the same total amount of area, then small patches must be important for diversity, right? While that can certainly be true, there’s two problems. First, we need to ask what difference in species number we should expect when we are comparing the number of species in habitat broken into different sized pieces. Is it reasonable to directly compare them in the way used in SLOSS analysis? Second, which species are accumulating so rapidly; are any extra species in the smaller patches just the most common species that you find everywhere, or do small and large patches contain the same mix of rare and common species? These questions were the focus of a recent paper of mine.

What should we expect to see?

Thinking about the first question (i.e., is this a reasonable comparison) the answer is a pretty clear no. For one thing, we know that sub-dividing habitat results in an expectation of more species in several small samples. The reasons are slightly technical, so I will focus instead on two other problems that arise from combining patches in reverse size order.

We know that species density (the number of species per area of habitat) decreases as you sample larger areas. The pattern we see (the number of species) changes depending on the lens through which we choose to view it (the size of the sample, or patch, or island etc.). This is an example of a phenomenon known as scale-dependence, and it is not best practice to ignore this in biodiversity comparisons because you are not comparing like with like. In SLOSS analysis, we completely ignore scale-dependence and all the highest species density (smallest), patches are combined first, with no effort made to account for how this might affect our expectation for their combined species number (I know of no other method in ecology where anything remotely like this is allowed).

Next, note that as we combine two patches, any species they share will not increase the total number of species. This becomes a problem here, because the number of species that are shared (proportionally) increases as the number of species in that patch increases. As smaller patches have fewer species, they will share fewer species (proportionally) than larger patches. This is quite intuitive when you think about it – there are only so many species in the landscape. Smaller patches contain a smaller fraction of them, so when we combine two of them, it leaves more species for the second patch to choose from without repeating the same species in the first. Put simply, large patches – just by chance – share a higher proportion of the species they contain with other large patches, smaller patches share a smaller proportion.

Now think about the SLOSS curves. In small-to-large order, the advantage of having more species per unit area in smaller patches (greater species density) compounds with every new patch we add to the curve. What’s more, of those species we do find, a smaller fraction of them are shared on average. What this means is, with no other information, the chance you encounter new species as you combine patches is greater for every patch in small-to-large order (Fig. 2). Contrast this with a typical sampling design, where all the patches are the same size – here the chance of encountering new species is the same for every patch and does not change as we combine them (‘Constant area’ in Fig. 2).

Figure 2. The chance that the species we encounter are new species is higher for every patch when we combine them in small-to-large order (open circles). If we are trying to ‘find’ all the species in the data, adding them in small-to-large order is going to maximise our chances for every patch we add. If all the patches were the same size, they would contain on average the same number of species. Then the probability does not change (diamonds).

So in SLOSS analysis the same area and number of species are being combined in each order of combination – but each species ‘counts’ for more toward the total in small-to-large order. This is the ‘green light route’ in my traffic analogy. It should not then surprise us much to find out that small-to-large SLOSS curves often rise very sharply initially, while large-to-small curves have a characteristic shallow slope. This explains why in about 70% of comparisons or more, small-large order seems to accumulate so many more species. If small patches were truly this important for species diversity, it should show up in all types of comparisons between small and large patches – it doesn’t. Instead there is massive debate about the importance of small patches.

Figure 3. Combining the smallest patches first gives every species in the patch a better chance of adding to the total number. Unless we account for this expectation using a null model (solid lines) we might conclude species just love to hang out in small patches.

Now, anyone who has done a few of these comparisons will say ‘hey, wait a minute – sometimes the curves overlap or the large-to-small curve is above the small-to-large one!’. Quite true and there are a few reasons – including it could be a real size-dependence pattern. The problem is we can’t tell from these plots. Who wins the race to capture all the species in the dataset for the smallest total area instead depends on another influence. To introduce the idea, I’ll return to my analogy: what if the ‘red lights route’ had no speed limit and you could travel as fast as you like? You might be able to overcome the disadvantage of the constant red lights and get to work quicker (yay). Back in SLOSS analysis world, the role of the speed limit is played by the island species area relationship.

The island species area relationship tells us about how the number of species we see changes in smaller or larger independent habitat patches. ‘Independent’ here roughly means isolated – the patches are far enough apart that the composition of one has little to no influence on the composition of those nearby. We can think of it as how a random sample from all the species that live in the region (the regional species pool) changes depending on the size of the sample we take (represented by the area of the patch). If the slope of the island species area relationship is shallow, there is not much difference between small and large patches and the ‘green light’ advantage means the small-to-large curve is likely to lie above the large-to-small curve. If the slope is high enough, large patches contain so many more species it doesn’t matter that small patches share a smaller fraction of species (because we can drive faster between lights).

Most power-law island species area relationships have a slope (aka z-value) of about 0.25 or less. Turns out we have to get to about 0.35 or greater before the large-to-small curve reliably gets it’s nose above the small-to-large curve. Because more datasets have a shallower slope, so more SLOSS analysis comparisons favour small-to-large order. This is what scientists call a ‘confounding factor’ – an effect that influences the outcome (the slope of the island species area relationship) but is not related to what we want to test (the effect of patch-size on species accumulation when combining patches). We might see a result, but we can’t say for certain why.

Figure 4. SLOSS analysis outcomes are strongly determined by the slope of the island species area relationship. Small-to-large lying above the large-to-small curve (as in Fig 1C) are favoured for slopes less than about 0.3. Data from 202 published datasets.

Are all patches created equal?

I mentioned I was interested in both the strength of evidence that the SLOSS analysis approach offers and the question of which species accumulate more rapidly in the smaller patches. I answered both of these questions using a null model. This is an approach where you create a bunch of simulated datasets that are like the one from the community you are analysing. However, you randomize how species are assigned to samples in each simulated dataset so you remove any structure in the dataset. Each of your simulated datasets is analysed the same way as in the observed data. From the datasets simulated under your null model, you know how the effect you are interested in would look like if it happened at random. That way if you see a bigger (or smaller) effect in the real data, then you have some evidence it would not happen at random. The null model I used predicted how many species we should see in groups of patches if patch size had no effect (you can see the null model in Fig. 2 – it is the smooth line). Here I randomised the patches that species were found in but kept the total number of times they were recorded the same.

Then, instead of comparing the small-to-large and large-to-small curves with each other, I compared each of them with themselves; that is, to their respective simulated curves. For every patch on each curve, I took the difference between the observed proportion of species and the expected (average value) from all the simulated communities. I then averaged these differences. If there were no effect, this value would be zero. If the average difference was positive, then when patches were accumulated in that particular order (either large-to-small or small-to-large), they contained more species than expected (negative meaning less). I interpreted a positive effect in large-to-small order (more species in the data than the expected value) as meaning that some species either preferred or required large habitat patches. Vice versa for small patches. I ran the analysis on 202 published datasets I have collected from the literature for three different habitat types: ‘true’ islands (i.e., archipelagoes), fragments of formerly continuous habitat, and ‘habitat’ islands – those which naturally occur as different size habitat patches in some non-habitat landscape (like wetlands, lakes and sky islands).

To explore patch size dependence for each dataset, I tested for a statistical difference between the small-to-large and large-to-small effect sizes. Over all 202 datasets I had available to run the model, the difference was -0.027. This I roughly equate to meaning for a given amount of habitat area accumulated, 2.7% of species would prefer you chose the largest patches. This value differed quite a lot between habitat types though (Fig. 5). It was strongest for archipelagos (~5% preference for larger islands), weaker for fragments (~1.5% preference for larger fragments) and there was no statistical evidence for any patch-size effect for habitat islands.

Figure 5. Patch size effects on species accumulation by habitat type (H= habitat islands, F = fragments, A = archipelagos). Effect size is the average difference from the null model for small patches minus the average difference for large patches. Thus, a negative value means more species prefer or depend upon larger patches and a positive value means more species prefer or depend upon small patches. In other words, the more density occurs to the left of the dashed vertical line at 0.0, the more important larger patches in that habitat type.

So, what do I conclude from all this? First, it is worth noting that the ecological effects of breaking habitat apart are much harder to detect than those due to the loss of habitat area (which is always bad news for biodiversity). Where present, the effects of breaking habitat apart are typically small in magnitude and can work in either direction. This is consistent with this analysis but contrasts with SLOSS analysis, which more often than not seems to imply small patches are collectively amazingly diverse. Despite the modest effect size with these data, this analysis suggests that small and large patches do not usually support all of the same species. It is likely that in most cases a subset of species require (or at least prefer) larger patches to persist.

Getting a little more specific, it was interesting to see the lack of clear evidence for large patch size dependence in species of habitat islands. I also note there were instances in all habitat types where a preference for small patches was evident. Unfortunately, the datasets where this was the case were not revealed through SLOSS analysis. In other words, the method doesn’t even reliably detect situations where small patches are preferred by more species than large patches after controlling for the problems I discuss above. On a personal note, after more than five years messing around with them in one form or another, I do not feel anything useful can be learned from SLOSS analysis about the effects of sub-dividing habitat (or anything else to be honest). Best avoided.

A ‘place for everyone’ – why are more species found in sub-divided habitat?

Predicting diversity in sub-divided habitat just might be ecology’s greatest theory-data fail

The question of why groups of small patches of habitat tend to contain more species than a single large patch of equal total area pretty much inspired my, recently completed, <sigh of relief> PhD research. I think I know exactly why – in fact, I think it’s almost inevitable. But I’ll save that for a later post (the answer’s in my thesis if you can’t wait). Here, I want to deconstruct the pattern a little and explore support for some possible explanations.

It is a deceptively simple question. One that has garnered some complicated explanations (that can come across a little like excuses) to account for some inconvenient evidence. When I say ‘inconvenient’ I mean contradictory. I actually think it could be one of the best examples of a head-on collision between well-validated predictive theory and sheer weight of empirical evidence in the whole of ecology. So what’s the problem? 

Those ‘well-validated theories’ are largely based on demographic models (which track births, deaths, immigration and emigration over time and space to predict changes in population size).  In particular, I’m thinking of island biogeography and metapopulation theory.  I’ve touched on the influence of island biogeography theory (IBT) on conservation thinking before. Metapopulation models also use demographic rates (births, deaths and so on) to predict how a population of one (two, a few) species is affected as these rates are varied within and among of a bunch of spatially separated (discrete) sub-populations. That is, in a ‘population of (sub)populations’ (= a metapopulation).

Cutting a long story short, demographic theories encourage us to expect small patches of habitat will contribute little to conservation for two main reasons. First, they will contain only a select group of only the most common (and presumably therefore, low conservation interest) species. But not only that – any sub-populations of species they do support are prone to go extinct anyway, because small patches can support only a few individuals and they might all die out just by chance. It is quite reasonable to conclude from these very large bodies of well-validated theory that a single large patch will support more species than a bunch of small patches.

However, lurking uncomfortably in the shadows is the weight of evidence from dozens of published Quinn-Harrison curves, where we usually find the exact opposite pattern – it is groups of small patches that seem to contain more species. Prof. Lenore Fahrig has been an almost lone voice in pointing out this disconnect between theoretical expectation and observation on the effects of habitat subdivision (or ‘fragmentation per se’) on diversity. On reviewing evidence from Quinn-Harrison curves, Fahrig found near-universal support for small patches containing more species.

More recently, Prof. Fahrig has gone on to ask why we see this pattern at all. I think that is the right question. In my view, reducing the ‘small patches have more species’ observation to  debate over reserve design skips a crucial step. Understanding. If we focus on understanding how the pattern arises, we might not only narrow the scope for debate, but – more importantly – we might even learn some very useful lessons about managing diversity in fragmented landscapes. 

So – why?

There are a few theories, but not a lot of tests. One suggestion is that the species concerned are only common, widespread species. Hence it is of no relevance for conservation if  small patches do have more species. I have found that when you look at the evidence there is not much to support this idea. Prof. Fahrig agrees. Still, it is probably the most common expectation among ecologists and to some extent that is reasonable. But a common-species-only argument does not answer our question.

The main theory to explain why groups of small patches contain more species than a large patch of equivalent area relates to environmental variability. Because subdividing habitat means each smaller area must be further from other small areas than if they were in a single patch, the small patches are likely to differ more in their biological and environmental conditions. So, there’s really two sides to this: biologically, you might be able to escape better competitors by just finding somewhere to live where those guys don’t; environmentally, you might have a better chance of finding the perfect set of conditions to support your growth.

I like to think of this as the ‘place for everyone’ hypothesis. And what we would expect to see is that the species found in each small patch is really quite different – at least more different than we would expect if we sampled areas of the same size from within a single large patch. This is essentially what Quinn-Harrison curves tell us – that for a given total area, species composition within and among large patches is less variable than species composition within and among small patches. 

Ecologists call change in species composition between local areas ‘beta diversity’. That this increase in beta diversity is due to environmental conditions is theory # 1 for why small patches have more species (note that this really largely ignores the ‘biological’ side of things).

Another thing that could result in a difference in species number between small and large patches is if they differ in the way that individuals are divided up among species – their abundances, relative to one another. In nature, a few species are common, and lots of species are pretty rare. We quantify these differences using a diversity property called ‘evenness’. If every species had the same number of individuals, a common measure of evenness (proposed by the awesome quantitative ecologist E.C. Pielou) would assign this a value of 1. But it is usually (read, ‘always’) the case that species differ in their number of individuals; a highly uneven distribution would have an evenness value close to 0. The figure below illustrates the basic idea.

Figure 1. Evenness of two bird assemblages (bird pix and loads of other stuff are freely available from the excellent IAN library – Integration and Application Network, University of Maryland Center for Environmental Science ian.umces.edu/imagelibrary/).

To show why evenness matters, let’s ask what might happen if two birds died from each of the assemblages in  figure 1. What effect would that have on the number of species? Well, for the panel on the left it could remove half the species if you were unlucky; but in the assemblage on the right, it could never remove any species completely.

Conversely, if you grabbed (say) three individuals at random, the right panel would, on average, yield more species. We can estimate this using a formula derived by Stuart Hurlbert. Here, if we grabbed 3 individuals at random from the left panel, we’d expect 1.74 species; from the right panel it is 2.5 species.

The take home is that both extinction risk, and the number of species we expect, depends not only on how many species there are, but also on how many of each species there are as well. This is true whether the habitat is sub-divided or not. If small and large patches have the same number of individuals, but the small patches have a more even distribution of abundance among species, then they should also contain more species.

This is illustrated in the next figure.

Figure 2. Conceptual diagram illustrating how differences in the evenness of abundance between small and large patches could affect the number of species they contain. Panel A shows two species abundance distributions, where each species is a symbol and the species are ordered from the most abundant (left) to the least abundant (right). This is their rank. An even abundance appears more horizontal (if all species had the same number of species, it would be a straight horizontal line in this plot). We can ask how many species we would see in several small patches, relative to a large patch of the same area, by simulating a species accumulation curve starting with the smallest patches and comparing it with a large patch (dashed lines in Panels B and C). Now, if the small patches assemble from the uneven distribution and the large patch from the even, the large patch will contain more species (panel B). If the opposite is true, the small patches will contain more species (panel C). Adapted from Deane et al (2020).

There’s a twist though: recall that we measure change in species composition using beta diversity? Well, this also depends on evenness – but in the opposite direction to species richness. So, samples from more uneven distributions will result in higher beta diversity. This is interesting because, as we discuss above, higher beta diversity is the main explanation for more species in those small patches.

So rather than asking about beta diversity, in some ways it makes more sense to ask whether and how species evenness changes for patches of different size. Because that will tell us about both the number of species and beta diversity. Typically we expect abundances in any single small patch will be very uneven. However, if evenness increases in groups of small patches (which some people have found), it should decrease beta diversity. The key point here, is that despite reducing beta diversity, increased evenness in small patches could still potentially result in more species in several small, than a single large patch (as Figure 2 shows).

There’s another possibility that has been sort of acknowledged in the literature but not directly tested: what if the pattern is only due to a sampling error? Think about how much easier it is to count the number of birds in a backyard compared to a large urban park. Maybe (probably) we do not count the species in the large patches as well as we do in the small patches. Because of that bias, we might miss more species in the large patch. So it might seem that the small patches have more species, because we just didn’t count the large patches properly. If you have the right kind of data, you can control for this (but that is not usually the type of data available to build Quinn-Harrison curves).

Finally, if we want to find a general explanation, we need to consider how possible differences among the different types of animals and plants will influence the patterns. For example, birds are very mobile, other animals not so much – would we expect to see different taxonomic groups like insects and mammals with the same pattern of richness? Some animals eat plants, others eat other animals, will they have similar area requirements? Clearly we need to account for this potential source of variation and test for differences among taxa as a sub-question of asking why small patches contain more species.

Recently, we combined all of the above ideas (beta diversity, evenness, sampling error and taxonomic group) into a single model – called a structural equation, or confirmatory path analysis model. We used the model to test how much influence each of these possibilities has on the difference in species number between a single large patch and a group of small patches. We used a bunch of published data to do this, but because our data had some nuances requiring special treatment we used a fairly new kind of path analysis called piecewise structural equation modelling (See here for some excellent background on the method, which is under ongoing development led by ecologist, Jon Lefchek).

The figure below sets out the hypothesized paths we tested and how they relate to one another. The thing we are interested in is the difference in species number between groups of small and a single large patch (the dashed line in Figure 2; ‘Difference in species number’ in Figure 3). If small patches have more species, this number is positive, if the large patch has more species, it’s negative.

What figure 3 essentially shows is how we imagine the different influences might affect the number of species in small vs. large patches. The cool thing about path models is they are graphically based with the ‘graph’ representing the relationships you are portraying. This forces you to think carefully about the causal influence of the different factors included in the model.

Each arrow is really a hypothesis; if the arrow ends at the ‘Difference in species number’ box, it is called a ‘direct’ effect. If the arrow points to something else, then that points to ‘Difference in species number’, we call that an indirect effect. An indirect effect will influence the difference in species number because it affects the something else and it is that something else that influences the outcome of interest. 

In words, we thought the difference in species number (our response variable) would be directly and positively affected by beta diversity (more beta diversity means a larger positive difference in species), directly and positively affected by evenness (more even abundance in small patches means more positive difference), directly affected by taxonomic group (more of a control than a hypothesis) and directly and negatively affected by sampling error (more sampling effort should decrease the difference in species number).

The main indirect effect we hypothesized was the negative influence of evenness on beta diversity, although we thought taxonomic group would probably also have an indirect effect, albeit we were not as certain what direction it might go.



After we tested the hypotheses (using data from 44 published studies), we found support for most of the paths. This is shown in the next figure, where we only draw an arrow where we found an effect (the hypothesis was supported). The size of the arrows roughly shows the relative strength of that effect, while the number beneath or above the arrow (the path coefficient) quantifies its strength, and shows the direction of the effect (positive or negative).


Results of path analysis (adapted from Deane et al 2020). Supported paths have an arrow and the effect size is proportional to the width of the arrow.

The first thing to note is that we were able to explain nearly 70% of the variation in species number between small and large patches using our causal model. In ecology, that’s pretty respectable, but it’s not a huge sample size, so nothing to get too caught up in. Next thing is that the direction of the effect for all direct and indirect hypotheses in the model were as we predicted, which is much more satisfying. Though a number of the paths (hypotheses) were not supported by the data (compare the missing arrows in the two path diagrams).

Also, while most of the variation was explained by differences among taxonomic groups, the main model predicts that we should, on average, always expect small patches to contain more species than a single large patch, irrespective of taxonomic group. Is it possible for a single large patch to have more species? Sure. But should we expect that for any taxonomic group based on this analysis. No. However, the sample size (44 is not many data points for an analysis like this – few researchers are generous enough to make this type of data freely available) means we need to be cautious in drawing any general conclusions.

So then, what’s the main take homes from this analysis? Taxonomy matters the most, then beta diversity. In other words, small patches matter for all animals, but some a lot more than others. After accounting for the type of animal, evenness really has a big effect on species number both indirectly – via effects on beta diversity – and directly where increasing evenness increases the Difference in species number. But because it is opposite in direction, the overall effect is quite weak.

What we can say, is the greatest difference in species number between groups of small patches and a single large patch occurs when both beta diversity and evenness are high. We should expect to see that when:

  • the taxonomic group we are interested in finely divides the environment into different niches – typically this will be small things like invertebrates (and also plants, which seem to respond in a similar way to invertebrates); and,
  • each patch presents contrasting environmental conditions – for example, where there are clear gradients in soil texture or nutrient concentrations; and,
  • it is not easy for individuals to move from one patch to another (lots of dispersal means that the same species that are the most prolific reproducers can reach every patch)

Basically the lower two points are conditions where the ‘place for everyone’ hypothesis should hold up well.

It is a pretty general set of conclusions, and quite what it means for managing landscapes with patchy habitats will depend on the circumstances and the type of animal or plant that you are interested in. But to me, it at least makes a lot of sense in explaining the high beta diversity I have seen in seasonal wetland plant communities (which vary greatly in hydrology, soil type, land use etc – see a Quinn-Harrison curve here) and also why their species composition becomes more similar when lots of those wetlands are found close together (because this increases dispersal). 

Do I think a group of small reserves will preserve more species than a single large reserve? No. Not really. But the landscapes I have worked in tend to be productive landscapes and the notion of designing a bespoke conservation reserve network is little more than a fantasy. In those landscapes, it gives me great hope that in at least some situations, groups of small patches do seem to provide a ‘place for everyone’. At least for now. The questions then become: how viable are those species’ populations; and, how can we best help them to persist as part of a functional, productive landscape.



When isn’t size everything?

Biodiversity is best protected in large, undisturbed areas; but it seems that not every species will be represented in the largest habitat patches in a landscape. Why?

* See also my essay in The Revelator

Dating back to the 1970s ecologists have been interested (which for scientists generally means disagreeing) in why groups of small patches of habitat often (in fact, nearly always) contain more species than a single large patch of equivalent area. The whole thing blew up when Jared Diamond published six principles for conservation reserve design, based on his research in marine archipelagos, the species-area relationship and MacArthur & Wilson’s theory of island biogeography.

Diamond’s first principle states: A large reserve is better than a small reserve (principle A), for two reasons: the large reserve can hold more species at equilibrium, and it will have lower extinction rates (Diamond 1975, Biol. Cons. 7:129). 

The second part, that extinction rates are lower in larger reserves is pretty uncontroversial even now. To be fair, the first part is also true if you compare only a single small site with a single large site – but that is really more a statement of a well known pattern in ecology known as the species-area relationship rather than a design principle. People soon began wondering if it that part about holding more species was true if multiple small patches of an equivalent area were compared with that single large patch. Probably not as it turns out – a number of authors, starting with Daniel Simberloff and Lawrence Abele writing in the journal Science, made their alternative views on this clear.

Simberloff and Abele suggested that the species-area relationship could be used to argue the opposite conclusion. In a back-of-the-envelope illustration, they showed that unless the large reserve contained nearly every species in the landscape, then two smaller patches of equal area will hold more species. And the more pieces you broke the area into, the more additional species they would collectively contain. These exchanges initiated a long-standing debate in conservation biology as to whether it was better to have a Single Large or Several Small reserves for species conservation. It is known as the SLOSS debate.

People then started getting interested in testing support for the two sides of the SLOSS argument with their data. One way you can do this (at least visually) is by plotting two curves on the same set of axes. Both curves show how many species you accumulate as you combine patches (the y-axis), for the amount of area found in the patches (the x-axis). The difference comes from the way the sites are combined – specifically the order. One curve is built by taking the patches in order of increasing size starting with the smallest, then adding the second smallest and so on until you have combined all of the patches. The other curve plots the same information, only it is calculated by combining patches from the largest to the smallest. For a given amount of accumulated area, the differences can be quite pronounced (as illustrated in the figure below for some wetland plant data from South Australia).

Comparison of the number of species accumulated when combining sites in order of their size. Starting with the smallest site and then adding the next smallest and so on, the blue curve results. Doing the opposite and starting with the largest and adding the second largest and so on, the salmon coloured curve results. For a given proportion of wetland habitat area accumulated, many more species are found in a collection of smaller sites than in the equivalent area spread among larger sites.  Data are vascular plants from 76 Fleurieu Peninsula wetlands and are freely available on the Aekos portal website.


James Quinn & Susan Harrison first introduced the method of comparing the two size-ordered curves and in the same paper analyzed 30 published datasets. They found that in 29 of these the small-to-large order saturated (reached the total number of species) before the large-to-small curve.

Apart from sparking an enduring debate about the relative virtues of single large vs. several small reserves (missing the point a little in my view), this pattern has generated rather little interest. I think there are two interesting aspects to it. First off: why? Why should the diversity of many plant and animal species consistently assemble this way in very different types of patchy island-like habitats?

Second, it is well known that smaller habitat patches in a landscape are more vulnerable to complete loss (this even has a name: ‘attrition’). For example, the smaller a patch is, the more vulnerable it is to extreme climatic cycles (which ponds or wetlands you would expect to dry out first in an extended drought?). They are also easier to clear and generally lack formal protection. The most likely outcome in a human-dominated landscape would be conversion to some human land use.

So perhaps the question we should be asking is how would the loss of these smaller patches affect diversity in that habitat type across the landscape?  If the rapid accumulation of species for a given amount of area is simply widespread generalist species, then loss of small patches would not matter – at least for species numbers. But if they contain some of the rarer species in the landscape, then species diversity could be impacted. This was exactly the question that inspired our recent paper in the journal Global Change Biology.

We first compiled 175 published datasets for all types of discrete habitats. By ‘discrete’ I mean patches of some sort of habitat that were surrounded by some sort of non-habitat. So we included wetlands, ponds and lakes, oceanic and nearshore island archipelagos and even sky islands formed by high-altitude zones of mountain peaks. We also included artificially discrete habitat types – that is remnant fragments of forest or grassland. And we included all types of plants and animals. The only real criterion I had was that the data were something close to a full census of all the species in each surveyed patch.

We then simulated the destruction of the smallest patches to a total of 10 or 20% of the total area and calculated whether any species would be lost from that network of patches. Over 80% of datasets indicated at least some species would indeed be removed, even if all of the largest patches and their species were preserved. That did not surprise me, but the proportion of species that would be lost was a little unsettling. Because even if those smallest patches destroyed represent only 10% of the total area contained in all the patches, on average between 7 and 9% of species would be lost.

That might not sound like much, but it is an outrageously high proportion. By way of comparison, the most widely used species-loss model based on reductions in area (the backwards species-area relationship, described here) would predict about 3% species loss for a 10% loss of area. And this model is prone to over-predict extinctions in continuous habitat areas.

For this result to arise, it must mean that rather a lot of species are found only in those smaller patches. For me, the most interesting thing about the analysis was comparing the patterns of species loss among the datasets and trying to figure out why that is. Although the size-ordered species loss curve for every dataset was unique, there were four clear types of patterns that they could be grouped into. These are interesting because they give us some hints about why it is so commonly the case that some species are found only in those smallest patches.

What happens to species diversity when you remove patches of habitat from the smallest to the largest? Each point shows how many species are lost from the network (on the y-axis) for the proportion of the total area you have accumulated (x-axis). The solid line shows how many species we should expect if species were randomly found in patches and the dashed lines show the uncertainty in that model. Although the shapes of the curves were different for every dataset they all fell into one of four different types of pattern. In panel (a) it does not matter how many small patches were destroyed, no species were lost. This was most common in mammals. In panel (b) species loss was random and this was common in highly disturbed habitats. In the step pattern (c), loss of most small patches does not matter, but occasionally, loss of a single patch leads to loss of species.  Plants and invertebrates lost more species due to their more segregated distributions, mainly following the step and d. linear pattern.

For larger animals we would expect that small patches are a poor to totally unviable habitat and only the most common, generalist species would be found in them. For those species, you could probably destroy all the small patches and still lose no species. And that is exactly what you see in what I called the threshold pattern (top left panel (a) of the figure). That accounted for the roughly 20% of datasets that lost no species and these were mostly vertebrates.

In highly disturbed situations, we might expect more random patterns of occupancy. In that case it is likely that a few species might happen to occur in only the smallest patches. Again, this is consistent with what I found in about 12% of datasets. These included temporary ponds, weeds in vacant lots and also migratory breeding birds that re-colonize each year. The common theme there seems to be regular, high levels of disturbance.

The most interesting and most common (~45% of datasets) pattern for me was what I called the step pattern (panel c, top right). Here you could ‘safely’ throw out most small patches without any species loss, but every now and then removing a patch led to the loss of one or more species. Like the threshold model, it featured a nearly horizontal line on the left side (indicating that few species were lost as small patches were destroyed) and only started to show rapid species loss above some threshold size. But here it was different, because some of those small patches did support species not found in larger patches.

I could imagine two situations where the step pattern could arise. First, maybe some of the small patches were unique in their environmental conditions – maybe a fresh groundwater spring in a largely saline landscape supporting locally rare, less salt-adapted species.

Second, if some species have a minimum patch size threshold, might not some species have a maximum size threshold – patches that are too big for them to live in? The number one reason I can think of for that has to do with predators. If the things that like to eat you need a certain size habitat to exist (say fish in a pond), you might intentionally choose to live in smaller habitats to avoid them. Bill Resetarits at the University of Mississippi has demonstrated exactly this type of pro-small-patch selection behaviour in aquatic beetles and mosquitos in pond mesocosms.

Finally, there were about 20% of datasets where almost every patch destroyed contained some species not found in other patches (lower right, panel d. I called this the linear pattern, but it was not always such a straight line as this dataset). The pattern was most common in more diverse taxa like insects and plants, which require only small areas (at least relative to what humans might call small) to obtain enough resources to live. So this pattern seems to be driven by the scale at which organisms experience environmental heterogeneity – that is, it relates to the breadth of the niche that organism inhabits and it’s energy requirements.

It’s worth considering possible reasons why this pattern might show up in data, but not accurately reflect on-ground reality. Sampling design is a prime candidate, where researchers might (intentionally or not) avoid sampling small patches they know (or believe) to be lacking in species of interest. This seems likely, but I have no way to test it. It is also possible that larger patches are harder to census and rare species were simply easier to spot in the smaller patches. I don’t think that is the case though because there was no evidence of any relationship between survey effort and species loss. Similarly, the proportion of species lost had nothing to do with the sizes of the patches – so if it is just incomplete sampling of larger patches, it is remarkably consistent across scales.

But taking it on face value, I got a few take homes from the work. On the down side, if we were to lose a lot of small patches over a short period of time because of climatic extremes, or if small patches were just being incrementally cleared over time, then it seems likely we will lose some species from that landscape as a result. There is probably rather little we can do about extreme climatic events, but we can educate, encourage and legislate to prevent on-going intentional clearance of natural habitat patches.

I prefer to view it as evidence that preserving every patch of habitat in the landscape is probably of more tangible biodiversity value than I at least had realized. Importantly, the question of who lives where is about habitat quality. And habitat quality is in the eye of the beholder. Size it seems can be a rather poor proxy for habitat quality. That means that even small and isolated patches of habitat could contain species not found anywhere else in that landscape at the moment.

Finally, it’s worth noting that managing a landscape for species diversity is not everyone’s cup of tea. And also that the big headlines on biodiversity loss typically refer to biodiverse hotspots and global, not local, species loss. Fair enough. But my view on that is if we are managing our landscapes sustainably, then most of the native species that are around now, will still be there in a few decades time. The number of individuals in each species will vary depending on seasonal and other cycles and local populations will inevitably blink in and out. But if every time we get around to counting, we find fewer species in the landscape than the last time? Well – it’s hard to be neutral about that.