Better SAFE than sorry

30 11 2011

Last day of November already – I am now convinced that my suspicions are correct: time is not constant and in fact accelerates as you age (in mathematical terms, a unit of time becomes a progressively smaller proportion of the time elapsed since your birth, so this makes sense). But, I digress…

This short post will act mostly as a spruik for my upcoming talk at the International Congress for Conservation Biology next week in Auckland (10.30 in New Zealand Room 2 on Friday, 9 December) entitled: Species Ability to Forestall Extinction (SAFE) index for IUCN Red Listed species. The post also sets a bit of the backdrop to this paper and why I think people might be interested in attending.

As regular readers of CB will know, we published a paper this year in Frontiers in Ecology and the Environment describing a relatively simple metric we called SAFE (Species Ability to Forestall Extinction) that could enhance the information provided by the IUCN Red List of Threatened Species for assessing relative extinction threat. I won’t go into all the detail here (you can read more about it in this previous post), but I do want to point out that it ended up being rather controversial.

The journal ended up delaying final publication because there were 3 groups who opposed the metric rather vehemently, including people who are very much in the conservation decision-making space and/or involved directly with the IUCN Red List. The journal ended up publishing our original paper, the 3 critiques, and our collective response in the same issue (you can read these here if you’re subscribed, or email me for a PDF reprint). Again, I won’t go into an detail here because our arguments are clearly outlined in the response.

What I do want to highlight is that even beyond the normal in-print tête-à-tête the original paper elicited, we were emailed by several people behind the critiques who were apparently unsatisfied with our response. We found this slightly odd, because many of the objections just kept getting re-raised. Of particular note were the accusations that:

  • We were somehow promulgating that conservation decisions should be made solely on the back of SAFE (we never said this, nor support such a crazy idea); we merely point out that SAFE provides information that the Red List categories do not.
  • The IUCN threshold abundance criteria (mostly inherent in Criteria C & D) are not arbitrary and in fact based in a sound, empirically derived manner:

Well, we had a look at this in the original paper by Georgina Mace and colleagues justifying the abundance thresholds. First, While some values seem to have some logical basis and are in fact EXACTLY in line with our previous work on minimum viable population size supporting the ‘thousands’ (median: 5000) mark (in other words, our detractors have, retro-actively, agreed with us), all other thresholds are based on a theoretical paper published by Russ Lande in 1993 entitled: Risks of population extinction from demographic and environmental stochasticity and random catastrophes (American Naturalist 142: 911–927). Now, Russ is a bit of a theoretical ecology god, and the paper is great; however, the practicality (and biological reality) of the thresholds he derives can be questioned. His theoretical estimates are based on an unstructured, exponential model with a ceiling carrying capacity; while certainly informative, relying on such a relatively simple model that really only describes the effects of demographic stochasticity (kicking in at very small population sizes) should have rung some alarm bells. On the contrary, our analyses examining fully standardised MVP estimates from a wide range of taxa and based on empirical measurements give a more conservatively based estimate of a generalisable extinction-risk threshold. Thus, we come back to our original point – pegging a species’ population size to a generalised target makes intuitive sense and goes well beyond the theoretical (and simplistic) abundance threshold criteria provided in the Red List (but see previous point about not relying solely on SAFE – ours is an added dimension, not a replacement).

  • Almost NO species are assessed in the Red List based on Category E alone (population viability analysis); SAFE does this implicitly.
  • If our detractors are in fact so against any threshold criteria, then we propose that the power brokers of IUCN Red List should do the right thing and dump categories D and E entirely (as well as components of C). We would advise against this, but the corner in which they appear to be painting themselves argues for this outcome.

Now, I do not want to give readers the impression that my invective is in any way a criticism of our detractors’ research. I am a HUGE fan of many of them (e.g., the legend of Mace, the brilliance of McCarthy, the numerical genius of Akçakaya), I have published with some of them, and I clearly think that much of their research is directly and indirectly responsible for reducing extinctions. However, I am increasingly surprised by the blockade they keep putting up against the elegant simplicity of SAFE and its potential applications.

So, back to the conference in Auckland next week. I will expand on some of these issues and present some more data supporting our case. I imagine a few stone throwers will be in the crowd, so question time should be entertaining ;-).

CJA Bradshaw



3 responses

28 01 2014
We’re sorry, but 50/500 is still too few |

[…] when we were attacked in several high-profile journals, it came as something of a surprise. The latest lashing came in the form of an entire Trends in […]


27 09 2013
Too small to avoid catastrophic biodiversity meltdown |

[…] size. We’ve certainly written about the former on many occasions (see here, here, here and here for our work on minimum viable population size), with the associated controversy it […]


2 12 2011
Michael McCarthy

Hi Corey,
I’m not sure your blog represents our concerns accurately. Any repetition about this is simply because you seem to not acknowledge some points of fact.
I’m going to concentrate on two main topics that we raised in our published response:
1) The SAFE index should not be used for prioritization (solely or otherwise); and
2) The basic construction of the index is not properly justified.
To the first of these topics, there is now a solid body of research that shows that prioritization should be based on the management objective (including the relative value of different species if appropriate), the efficiency with which spending resources on different species meets that objective, the uncertainties in the measures of efficiency, and the available budget.
Some key references about this are (email me if you need copies:
Joseph et al. (2009):
McCarthy et al. 2008:
McCarthy et al. 2010:

The prioritization is derived by finding the investments in each species that optimize the management objective. That is, one finds the combination of conservation actions that minimizes the number of extinct species, or minimizes the number of threatened species, or best achieves whatever the particular objective is.
The SAFE index does not seem to apply to any of these aspects, so it is unclear how it has any role in prioritization. In this light, statements such as “Practitioners of conservation triage may want to prioritize resources on the Sumatran rhinoceros instead of the Javan rhinoceros (–1.36 versus –2.10, respectively).” from your original SAFE index paper and other statements about prioritization in the media (e.g., are not correct – the SAFE index simply does not enter into the prioritization equation.
To our second point, the claim in your response that the logarithm is “used for standardization purposes” is factually incorrect – the logarithm has no role in standardization.
The SAFE index is simply a measure of the difference between N and MVP. That is an elementary concept, so the need for the index is unclear.
But assuming we do want to combine N and MVP into a single value, you have pointed out that you want an index that is positive when N is greater than MVP and negative when it is less. The SAFE index is log(N) – log(MVP), but any monotonic increasing function will provide these properties; the log transformation is not necessary. The simplest index to satisfy the properties you seem to want is N – MVP. After pointing this out to you in our email correspondence, you responded:
“How hard is it to understand our log-transformation produces an intuitive result (i.e., that negative = bad; positive = good)?”
But I had already shown you that it was not the logarithm that gives this intuitive result. If there is any “vehemence”, as you put it, it is because you do not seem to acknowledge the veracity of this and other simple facts. I think it is more exasperation than vehemence. One possible response to your above comment could be:
“How hard is it to understand that the subtraction sign produces the intuitive result (i.e., that negative = bad; positive = good)? It is not the logarithm.”
It is possible that you use the logarithm because you want to emphasize proportional differences between N and MVP (the SAFE index equals log(N/MVP), so it is based on the ratio of the two numbers). But logarithms are not intuitive to most people. For example, how many people will know intuitively that a SAFE index of 0.301 means that the population size is double the MVP?
If you wanted proportional differences, isn’t (N-MVP)/MVP much more readily interpreted by the “general public, conservation donors, and policy makers”? In this case, the index would equal +0.5 when the population size was 50% higher than the MVP. And it would equal –0.2 when N was 20% below the MVP. That seems more intuitive.
But even so, if N and MVP were both reported (separately), I’m reasonably sure that people could figure out how far N is from MVP, In fact, I’d bet that most people will have an easier time figuring out the difference directly (and the implications of that difference) rather than using the SAFE index.
You refer to a “blockade”, which might create the impression of a coordinated campaign. That is not the case. The concerns expressed by others have been generated independently. Independent concerns also exist beyond the field of conservation biology ( My concerns largely stem from the errors of fact in your article and in subsequent correspondence, and how others might interpret the use of the SAFE index for prioritization.
Given that you refer to our email correspondence, and for the sake of providing context to your readers, I suggest that you publish all the email correspondence about the SAFE index between you and me on your website. Readers will then be able to judge questions of vehemence for themselves. I believe that I have interacted with you on this topic in good faith, and with honesty and transparency. And without vehemence.
By the way, I won’t be throwing any stones after your talk at the conference as you suggest I might. Question time isn’t the appropriate place given the couple of minutes that would be available for discussion. However, I’d be happy to continue the dialogue via your blog and face-to-face.
Michael McCarthy
School of Botany
The University of Melbourne


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: