Home
Mission
Overview of Project
Project Staff
Sponsors
Achievements
Checking, Illustrations
Upcoming Activities
Id and Species Lists
Protea Information
Protea Gallery
Growing Proteas
Interim Dist. Maps
Publications
Afrikaanse Inligting
Discussion on Red Data List Status
Below are is an academic spat in the form of e-mails, between Tony Rebelo and Steve Richardson
From: Tony Rebelo (15 September 2003)
To: Steve Richardson
According to the IUCN document [p.12], area of occupancy is defined as
the area WITHIN its 'extent of occurrence', ie occupancy [km2] should
always be less than or equal to occurrence [km2]. What you've got is a
scaling problem at the low end of the occupancy parameter, ie the
occupancy for LDCLIM and VELATE [3 km2] is exaggerated and should be
less than or equal to the occurrence [1 km2]. Otherwise, the algorithm
is an ass.
***Yes, it says so in the text, but not in the illustration. Note that the
cells for the outlying points are outside of the extent of occurrence.
I would go so far as to argue that any species which has an extent of occurrence
less than occupancy, where occupancy is at a scale of valid populational extent
in terms of gene flow and seed dispsersal, should be CR.
Of course, this makes no difference whatsoever to the reasoning for
LDCLIM and VELATE and I'm still happy they are CR. What's more
interesting is that the only way to 'save' these species [and get them
off my CR list] is by adding another location. This seems unlikely for
VELATE but is still a distinct possibility for LDCLIM! Is the burnt area
to the E of the existing location old enough yet for us to find one?
*** One locality will not do! The cutoff for CR is 10km2 which is 3 grid
cells. So you will need 3 populations at least 1.5 km apart from each
other. But even then, they will have to be 10km apart (at right angles) to
satisfy the occurrence criteria.
Regarding MIPALU, yes, the 60% Island code usage does look suspect, not
to mention the numerical coincidence(?) in Atlas plots conserved [59%]
and trashed [59%]. Beyond that you've lost me and I need to plead for
more background information on your 'decline' commentary. How and where
is TRASHED defined and how are the "59% trashed" and "21% trashed
in
occurrence and 19.5% trashed in occupancy" numbers derived? I'll reserve
judgement on the reasoning until I've caught up on that.
*** I checked up on the numerical coincidence when I saw it! The figures
are:
Occurrence: 28 (27.52) km2, Occupancy: 28 (28.34) km2,
Proportion of Protea Atlas plots conserved = 59% (10/17) with 59% (10/17)
trashed
And the near coincidence.
Conservation status - 39% (38.6) of occurrence and 40% (40.4) of occupancy
conserved
Trashed was defined as:
1. trashed layer as proportion of occurrence: OCCURRENCE
2. trashed layer as proportion of occupancy: OCCUPANCY
3. trashed layer as proportion of protea atlas points: PLOTS
4. protea atlas record localities with I, V, P, C codes: SITES TRANSFORMED
These of course give different values!
As for LSCONOC, I again need help as to how the TRASHED numbers are
derived though I'm happy the decline is significant. Rather I'd like to
extend the commentary on indicator and flagship species. In my view,
LDARGE is the ultimate flagship species for Peninsula Granite Fynbos
[including Orangekloof, Constantia, Wynberg Hill, upper Kirstenbosch]
BUT is only EN and not CR because of its exaggerated occurrence [1315
km2] as compared to LDCONOC [65 km2]. If you ignore the suspect inland
sites and simply use the occupancy figure [97 km2] as an estimate of
Peninsula occurrence, it would squeak in under the CR occurrence limit
of 100 km2 and qualify as
LDARGE CR, B1b(i+ii)c(iv) Occurrence: 97 km2, and declines in area and
fire-related population fluctuations
***Ah! But I need better evidence that the off Peninsula populations are
in fact man-made. Until then, Ld argenteum is not a Peninsula Granite
Fynbos endemic, but also occurs in Boland Granite Fynbos. But Ld argenteum
will be a good flagship species for Granite Fynbos (together with the Boland
Granite endemics Leucospermum grandiflorum, L. guenzii, L. lineare etc. Ld
daphnoides, Se kraussi (but some of the latter also occur on shale fynbos)
Then you would have both LDCOCOC and LDARGE as CR indicators of dry and
wet Peninsula Granite Fynbos respectively. Either way, both are doing
well in my 'mountain gardening' patch on the south side of Lions Head.
How's that for a pre-weekend riposte?
**Excellent! Keep up the conservation gardening!
Steve
Tony
From: Tony Rebelo (16 September 2003)
To: Steve Richardson
Hi
I am glad to see that on Tuesday I can face and rebut Monday's comments.
My comments >>>>>
Tony
>>> Stephen H Richardson <shr@geology.uct.ac.za>
15/09/03 20:26:50 >>>
Tony, now that we've narrowed down the list, here are a few more
comments I can't resist!
Key to cut and paste comments:
***=Tony
###=Steve
*** Regarding "area of occupancy is defined as the area WITHIN its
extent of occurrence", yes, it says so in the text, but not in the
illustration.
### Ah! BUT the illustration is merely an example. If you choose a grid
scale appropriate to the size of the occurrence, occupancy will always
be less than or equal to occurrence.
>>>>>But that is not the purpose of the occupancy. The
scale of grid size in occupancy is determined by the scale of the primary
threats. For aliens it is of the order of 1-10km per fire cycle, and for
agriculture about the same (the size of a field or orchard). It is also
determined by relevant biological aspects of the taxon: examples would be: size
of fires, distance of pollen and seed dispersal, distance moved by pathogens,
predators, scale of local disturbances (landslips, flooding), and local
vegetation patterns (seeps, rocky outcrops, deep sand traps). Thus our
scale of 1'X1' is a bit fine <although some processes are finer>, but
given our data <another criterion - we could go finer, but then we could not
include any modelling we are doing at the same scale> and given the localized
nature of some species populations, this is appropriate. {The Pretoria Red Data
Book team are happy with a 10kmX10km scale, but that is too course for my liking
- although it does eliminate problems with occupancy as nothing qualifies!} It
is thus often going to happen that very rare species will occur at a scale far
finer than landscape processes or threats. In such cases, occupancy will
be far larger than occurrance.
*** I would go so far as to argue that any species which has an extent
of occurrence less than occupancy, where occupancy is at a scale of
valid populational extent in terms of gene flow and seed dispsersal,
should be CR.
### I AGREE but a 3 km2 occupancy figure relative to a 1 km2 occurrence
is still an artefact of grid scale!
>>>>>I AGREE, but occupancy is interested in threats and biology
which may operate at scales much larger than a population, which in the case of
a very rare species may be the only population (nb population here is biological
not IUCN). The two measures are independent. I guess occurrence measures
the scale over which the species operates as a function of its history and
habitat. When the two measures approximate one another you have a very
rare and naturally threatened species.
*** One locality will not do!
### Oh yes, it WILL!
>>>>No it won't for occupancy: see below.
*** The cutoff for CR is 10km2 which is 3 grid cells. So you will need
3 populations at least 1.5 km apart from each other.
### FALSE. The CR occupancy criterion wouldn't apply since the a in
B1ac(iv) would no longer be satisfied.
>>>>Ah, but then it would slip in under B1ac(iv) - where a =
severely fragmented!
*** But even then, they will have to be 10km apart (at right angles) to
satisfy the occurrence criteria.
### FALSE. the CR occurrence criterion wouldn't apply since the a in
B2ac(iv) would no longer be satisfied.
>>>>>Same urgement as above. Two populations 10km apart is
about a Severe as is possible, other than of course, 100km apart. Note
that even 5 populations could still be severely fragmented and thus qualify!
*** The MIPALU figures are:
Occurrence: 28 (27.52) km2, Occupancy: 28 (28.34) km2,
Proportion of Protea Atlas plots conserved = 59% (10/17) with 59%
(10/17) trashed
Conservation status - 39% (38.6) of occurrence and 40% (40.4) of
occupancy conserved
Trashed was defined as:
1. trashed layer as proportion of occurrence: OCCURRENCE
2. trashed layer as proportion of occupancy: OCCUPANCY
3. trashed layer as proportion of protea atlas points: PLOTS
4. protea atlas record localities with I, V, P, C codes: SITES
TRANSFORMED
These of course give different values!
### Yes, but what are the TRASHED LAYER figures for 1. and 2. and how do
they relate to the Atlas data above?
And how does the value for 4. relate to "Major threats [for MIPALU]: 0%
of sites are transformed (agroforestry or urbanization)" in your
previous commentary?
Finally, whose I codes are these anyway?
OK - I had assumed that you would be happy with generalities. My apologies:
Trashed layer = layer used in CAPE, supplied by FRD, a landsat based analysis of
all transformed land in South Africa (we only used it for CFR taxa). Quite
good in Fynbos and Renosterveld (but mistakes irrigated pastures as natural veld),
but useless in Thicket and Limestone Fynbos (where Cynodon lawns in old fields
and pastures are mapped as natural) and not so hot oon the arid fringe where
ground cover is too low to distinguish vegetated and unvegetated areas.
This is very fine scale (too fine to believe I am guessing - 50mX50m pixels,
smoothed). To get trashed this layer is "intersected" with
occurrence and occupancy and the relative areas of trashed/untrashed extracted.
For PAP data, the localities are treated as points and the proportion of points
in/out this layer is the "Plots" estimate.
Protea Atlas record localities are based on the Vegetation Island Code on the
SRS: I, V, P, C = transformed, L,E,F = natural (blank=dont know).
expressed as a % {There is an error in the large MS I sent in that blanks were
counted as transformed, but the programme has been fixed now.}
Hope this clears this up!
What else would you expect from me on a Monday?
>>>>I hope you won't be too much sharper on a Tuesday!
Steve
From: Steve Richardson
To: Tony Rebelo (16 September 2003)
Tony, OK you've convinced me. This leaves only the trashed-layer-based
CR decline of MIPALU in doubt but a list of one is no longer a list.
List dismissed!
Yes, it's Tuesday - one week on and I'm ready to throw in the towel.
Many thanks for the magnificent duel!
Steve
From: Tony Rebelo (17 September 2003)
To: Steve Richardson
Hi
Easy! My computer does not know about fragmentation, I have not told
it about it yet. We were arguing about this for Ve late and Ld clim, so Ld
como h did not come into the story.
More technically, the computer also does not know that the one population of Ld
como h has not been seen since the fire, and that there is also a third
population of Ld como h waiting for us to go and atlas it. But I would
knock Ld como h up to CR over and above the computer listing for the former
reason, but leave it as EN if the populations as yet unatlassed were healthy and
large. Based on current knowledge, I would favour CR.
As for pistols at dawn? Nonsense! I would favour a good sceptre,
although for knock out power I would throw moderation to the wind and resourt to
immodoratum - although comosum would do quite well too!
Ta
Tony
>>> Stephen H Richardson <shr@geology.uct.ac.za>
17/09/03 09:50:59 >>>
Hey Tony, yes it's Wednesday and you haven't yet accepted my concession.
So there's still time for a new list to rise from the ashes of the old!
But not too long or that might reveal tactics. With MIPALU as a nucleus,
let's just add LDCOMOH in place of LDCLIM. Given your interpretation of
subcriterion a, if one extra locality would not do for LDCLIM [to get it
out of the CR category], how come LDCOMOH with occurrence and occupancy
well under the CR limits BUT 2 locations is only EN?
LDCOMOH EN, B1ac(iv)+B2ac(iv) Occurrence: 34 km2, Occupancy: 6 km2, and
locations =2 grid cells, and fire-related population fluctuations
Got to go
Steve
From: Steve Richardson
To: Tony Rebelo (17 September 2003)
***=Tony: My computer does not know about fragmentation...
###=Steve: It will after I've had a go at it with my tree popper...
See you in the silver tree patch at dawn!