ddi12369-sup-0022-Suppinfo

advertisement
1
(A) Supplementary materials and methods
2
(B) Range modifications
3
The range modifications can be grouped into eight categories of decreasing certainty: 1)
4
range reductions for species with recent anthropogenic human-induced range expansions
5
(24 species; detailed description in can be found Supplementary Data 1; graphical
6
representation of results can be found in Supplementary Data 2), 2) range expansions
7
based on other sources for species with known range declines (162 species; detailed
8
description in can be found Supplementary Data 1; graphical representation of results can
9
be found in Supplementary Data 3), 3) merger of likely anthropogenic disjunct ranges by
10
filling intervening suitable habitats (272 species; detailed description in can be found
11
Supplementary Data 1; graphical representation of results can be found in Supplementary
12
Data 4), 4) expansion of ranges to entire islands (198 species; detailed description in can
13
be found Supplementary Data 1; graphical representation of results can be found in
14
Supplementary Data 5), 5) expansion of ranges for species with known, or at least highly
15
suspected, range declines to cover suitable areas contiguous with the current range (231
16
species; detailed description in can be found Supplementary Data 1; graphical
17
representation of results can be found in Supplementary Data 6), 6) estimation of ranges
18
based on the natural ranges of the extant species that the target species co-occurred with
19
at fossil sites (186 species; detailed description in can be found Supplementary Data 1;
20
graphical representation of results can be found in Supplementary Data 7), 7) estimation
21
of ranges based on the natural ranges of the extinct species that the target species co-
22
occurred with at fossil sites (3 species; detailed description in can be found
23
Supplementary Data 1; graphical representation of results can be found in Supplementary
24
Data 7), and 8) unique species-specific modifications in special cases (13 species;
25
detailed description in can be found Supplementary Data 1; graphical representation of
26
results can be found in Supplementary Data 8). For many species, the range
27
modifications involved several of the above processes and we listed them under the type
28
with the perceived lowest certainty.
29
1) Range restrictions
30
The distribution was restricted for a few species with recent large range expansions
31
linked to anthropogenic range restrictions of other species, which is likely the case for the
32
expansion of Canis latrans (coyote) following the range decreases in Canis lupus (grey
33
wolf) (Arjo & Pletscher, 1999). The distribution was also restricted for species for which
34
colonization of a large part of their current distribution appears to be strongly connected
35
to human migration and land use, e.g., Mus musculus (house mouse) (Cucchi et al.,
36
2005). In addition, we interpreted the sudden appearance of commensal species on
37
isolated islands, contemporary with or after human arrival (Bover & Alcover, 2008) as
38
evidence of human-induced transport and removed these islands from the natural
39
distribution of such species. In total, we reduced the distribution of 24 species, from a
40
total of 36,421 occurrences to 24,628 occurrences in 110 × 110 km cells. Three species
41
(Canis latrans, Mus musculus, and Rattus norvegicus (brown rat)) had their ranges
42
restricted by more than 1000 occurrences.
43
2) Historical ranges
44
For historical ranges, some of the information, generally from areas with recent range
45
collapses, came from accurate sources such as Hall (1981), which covers North America,
46
but other sources, often those dealing with older range collapses, are more uncertain.
47
Despite the uncertainty of some of these sources, we are convinced that the estimated
48
ranges are closer to their natural ranges than the IUCN ranges. We changed the
49
distributions based on suggested historical distributions for 168 species, leading to an
50
increase from a total of 56,760 occurrences to 114,326 occurrences in 110 × 110 km
51
cells. A single species, Ceratotherium simum (white rhino), had its apparent range
52
restricted by 705 cells because the IUCN distribution simply maps entire countries
53
regardless of where in the country the animal occurs, whereas 13 species had their range
54
expanded by more than 1000 cells, Acinonyx jubatus (cheetah), Bison bison (American
55
bison), Bos primigenius (auroch), Caracal caracal (caracal), Camelus dromedarius
56
(dromedary), Felis silvestris (wild cat), Gulo gulo), Homo "Denisovans" (Denisovan
57
humans), Homo neanderthalensis (Neanderthal humans), Loxodonta africana (African
58
elephant), Lycaon pictus (African wild dog), Panthera leo (lion), Panthera onca (jaguar),
59
and Panthera tigris (tiger).
60
Modern humans (Homo sapiens), which are not mapped by IUCN and, therefore,
61
included in this list, also have an estimated historical range (Continental Africa) greater
62
than 1000 cells. Natural distributions can be defined for the other species in the genus
63
Homo, but the concept is less obviously applicable to our own species. However, the
64
hominin fossil record provides little indication of a stable co-occurrence of different
65
species in the genus Homo, even if hybridization between them has occurred to a limited
66
extent (Stringer, 2012). Continental Africa is the center of origin for Homo sapiens, and
67
no other species in the genus is known to have originated from the region during the Late
68
Pleistocene. Therefore, we assigned it as the natural range of Homo sapiens, as it can be
69
seen as the distribution of Homo sapiens that should have been maintained to avoid major
70
displacements of other Homo species during the Late Pleistocene.
71
3) Disjunct ranges
72
Disjunct ranges are occasionally a natural phenomenon, but many contemporary disjunct
73
distributions are anthropogenic in nature. This is evident for species for which
74
information is available regarding historic distribution, such as the historic range for the
75
larger African carnivores (Justina et al., 2005), but it is likely also the case for the many
76
species for which information on historic distributions is not available. If any of the
77
species we investigated had disjunct ranges, we looked for evidence or strong indications
78
of anthropogenic range decline (e.g., red-listing as endangered or critically endangered)
79
or evidence of poor knowledge of the true range (listed as “data deficient” or a note in the
80
IUCN description of poor knowledge of the species range). In cases in which such
81
evidence was found, but where we could not find information on the historic distribution,
82
we tried to connect the disjunct parts of the ranges unless we found evidence of
83
biological reasons for the disjunction. In particular, we did not expand ranges across
84
areas occupied by congeneric species that are currently not sympatric with the species in
85
question, as in the case of Hylobates lar (lar gibbon) in the Malay Peninsula north and
86
south of a thin band occupied by Hylobates agilis (agile gibbon).
87
For species for which the current distribution is fairly large (203 out of 272 species we
88
modified), we estimated the climatic convex hull using the climatic parameters yearly
89
temperature, winter temperature, and yearly precipitation (Hijmans et al., 2005) taken
90
from a rasterization of the range with grid cells of 0.1° × 0.1°. The convex hull was
91
estimated in R using the function inhull and the library geometry (Jewell, 2009; Barber et
92
al., 2013). After the convex hull was calculated, the disjunct ranges were joined if the
93
intermediate area was also within the climatically defined hull. This approach does not
94
guarantee that the intermediate area is suitable, but only that if the range is limited by our
95
chosen climatic predictors (or other factors with a sufficiently high correlation to the
96
chosen climatic parameters), the intermediate area should also be climatically suitable.
97
This approach implicitly assumes that the species is climatically limited; therefore, it
98
makes little sense for very narrowly distributed species. For narrowly distributed species,
99
we joined the IUCN ranges as lines or geographic convex hulls, when the intervening
100
areas are in the same habitat type as defined by the WWF ecoregions (Olson, 2001).
101
When the intervening areas consist of another habitat type (i.e. species occurring on both
102
sides of an alpine region) we joined the ranges as the shortest lines within habitat types
103
similar to the known occurences. In total, we joined disjunct populations in 272 species,
104
which led to an increase from a total of 18,113 occurrences to 23,882 occurrences in 110
105
× 110 km cells. The largest change was 558 cells for Felis margarita (sand cat), and in 70
106
species the changes were too small to be noticed on a 110 × 110 km grid.
107
4) Island species
108
Species historically endemic to islands smaller than 150,000 km2 (Java and all islands
109
smaller than this) for which a range decline was either known (including extinct species)
110
or at least very likely had their ranges expanded to the entire island(s). The current ranges
111
of many of the extant species in this category are ecologically narrow and often limited to
112
mountain forests. We acknowledge that it could mean true ecological specialization for
113
some of the species, but for most of these islands the mountains are the only areas with
114
natural habitat and, therefore, the only place where formerly generalist species survive, as
115
illustrated by the current distribution of many birds on Hawaii (Carter et al., 2009).
116
Despite the small potential exaggeration of the range by this expansion we think that the
117
estimated ranges are going to be closer to natural distributions than assuming that these
118
species are all mountain endemics. It should also be noted that all analyses were
119
performed on 110 × 110 km grid cells, so any exaggeration would only affect to the few
120
cells exclusively containing lowland areas. We changed the distribution this way for a
121
total of 198 species, leading to an increase from a total of 532 occurrences to 2,151
122
occurrences in 110 × 110 km cells. The largest change was 32 cells for four extinct
123
Caribbean species that expanded to all of Cuba and Isla de la Juventud: Boromys offella
124
(Oriente cave rat), Boromys torrei (Torres cave rat), Nesophontes major (greater Cuban
125
nesophontes), and Nesophontes micrus (western Cuban nesophontes). In 11 species the
126
changes were too small to be noticed on a 110 × 110 km grid.
127
5) Range expansions
128
A number of species have known, or at least likely, range declines, but with the historical
129
range either fully unknown or only known in parts of its range. Notable examples with
130
uncertainty in parts of the range include Ursus arctos (brown bear); Hall4 lists the
131
historical North American distribution, but the range decline in Eurasia is too ancient for
132
knowledge of the natural distribution to be known. A notable example of complete
133
uncertainty is Cervus nippon (sika deer), whose current range is very fragmented and
134
substantially smaller than the historical range, though the natural limits of the natural
135
range are uncertain. For such species, we estimated the convex hull and expanded the
136
range to suitable areas of either selected parts of the range or the entire range. Generally,
137
we only expanded the range to areas continuous with current ranges, but for some species
138
we increased the range to all suitable ranges within defined areas, such as large islands
139
(e.g., Madagascar) or areas with a known Late Pleistocene presence for some Australian
140
species (Supplementary Data 1). The change in the distribution of the 231 species led to
141
an increase from a total of 57,438 occurrences to 100,273 occurrences in 110 × 110 km
142
cells. Thirteen species had their range expanded by more than 1000 cells: Bison bonasus
143
(European bison), Canis lupus, Castor fiber (Eurasian beaver), Cervus elaphus (red deer),
144
Crocuta crocuta (spotted hyena), Cuon alpinus (dhole), Equus hemionus (onager),
145
Hippopotamus amphibius (hippopotamus), Lynx lynx (Eurasian lynx), Martes zibellina
146
(sable), Panthera pardus (leopard), Rangifer tarandus (reindeer), and Ursus arctos). In
147
all, but three of the 15 cases (Castor fiber, Cuon alpinus, and Equus hemionus), the range
148
expansions for these species were based partly on expertly drawn maps of historic
149
distribution. In 14 species the changes were too small to be noticed on a 110 × 110 km
150
grid.
151
6 and 7) Estimation based on fossil co-occurrence
152
For a large number of species, most of them extinct and a few surviving in very small
153
areas (e.g., Hypogeomys antimena, Madagascar giant jumping rat), we estimated
154
potential distributions based on extant species they co-occurred with in fossil or subfossil
155
species assemblages. The approach is similar to the co-existence approach for inferring
156
paleoclimate based on co-occurring taxa (Mosbrugger & Utescher, 1997). The logic
157
behind this is that if all species are limited by similar ecological parameters, the current
158
distribution of extant species should be informative of the climatic suitability of the area
159
for the species whose distribution we try to infer. We assumed that the frequency of
160
species from the fossil sites occurring at a given locality today corresponds to the
161
probability that the extinct species occurred there, which logically must be true on
162
average for all extant species. We assigned all grid cells with at least 50% of the species
163
from the fossil site as the likely current range of the extinct species, i.e., as presence.
164
Whenever possible we only used fossil sites with at least 10 fossil species. For species
165
with a large number of fossil sites, we used the single fossil site where the species in
166
question has been found with the highest number of extant species in each Taxonomic
167
Databases Working Group (TDWG) level 3 regions, generally corresponding to
168
countries, but with some large countries (e.g., Canada or USA) broken into states
169
(Brummitt, 2001). For some of the more climatically variable TDWG regions, such as
170
Madagascar and California, we used several widely separated sites if possible. These
171
rules were relaxed for many species due to a lack of available fossil sites, and many
172
species had ranges estimated based on fossils from only some of the TDWG countries it
173
used to occur in and/or based on fossil sites with less than 10 co-occurring extant species.
174
Most of the fossil data were based on Graham & Lundelius (2010) for Canada and USA,
175
Ferrusquía-Villafranca et al. (2010) for Mexico, and the Paleobiology Database (2015)
176
for remaining areas, but were supplemented by primary literature for species with
177
insufficient records in these sources. When several fossil sites were used, we used the
178
geographically closest fossil site to estimate the probability of occurrence in each grid
179
cell. When the resulting ranges from this analysis were disjunct, we joined them based on
180
climatic convex hulls as explained earlier, and a few species had their ranges expanded
181
based on the climatic convex hull, such as when historic occurrence was known from a
182
region where we could not find any multispecies fossil site. For closely related species
183
not known to co-occur historically, we restricted the ranges so they did not co-occur in
184
our estimated potential ranges. For this analysis, we used the inferred potential ranges of
185
all species rather than the current IUCN ranges. For three species, we could not find
186
sufficient data on co-occurrence with extant species and estimated their potential range
187
based on the estimated ranges of other extinct species, though this led to additional
188
uncertainty.
189
Our motivation for choosing this approach is that it enables us to estimate the potential
190
distribution based on a few (or only one) specimens, which is important because a large
191
proportion of the species we wish to map are only known from a few fossil findings. This
192
is most obvious for Brachyprotoma obtusata (a giant skunk), for which FaunMaps’
193
massive collection of fossil information only has three known Late Pleistocene fossil
194
occurrences from Pennsylvania, Utah, and Yukon (Graham & Lundelius, 2010). An
195
advantage of our approach is that an increase in the number of fossil sites for a species is
196
not guaranteed to increase its distribution, but may increase the precision of the
197
inferences. Thus, large and relatively easily identifiable species, such as Mammut
198
americanum (American mastodon), will not necessarily have larger inferred ranges than
199
species such as the aforementioned giant skunk that, for whatever reason, occurs rarely at
200
fossil sites despite apparently having a wide distribution. A problem with our approach is
201
that we assume that all fossils from a site are contemporary and, though we only use
202
fossils from a single fossil layer when such information is readily available from the
203
sources, the assumption of contemporaneity is unlikely to be entirely valid. Therefore, we
204
may potentially estimate distributions based on assemblages of species that have never
205
co-occurred, which could bias the estimated ranges. Still, the bias would normally lead to
206
a reduction in the size of the inferred range because species that did not overlap in time at
207
the fossil site would also be less likely to co-occur today.
208
Some fossil sites, especially in Europe, were dominated by hyper-generalists, which are
209
not informative of the climate of the fossil site; therefore, we removed them from species
210
lists prior to analyses of co-occurrence. In order to define hyper-generalists, we separated
211
the world into 10 habitat types, eight of them identical to the biomes of Olson (2010): 1)
212
Tropical and Subtropical Moist Broadleaf Forests, 2) Boreal Forest/Taiga, 3) Tropical
213
and Subtropical Grasslands, Savannas, and Scrublands, 4) Temperate Grasslands,
214
Savannas, and Scrublands, 5) Tundra, 6) Montane Grasslands and Scrublands, 7)
215
Mediterranean Forests, Woodlands, and Scrublands, 8) Deserts and Xeric Scrublands.
216
The remaining habitats were made by merging biomes from Olson (2010). 9) Tropical
217
and Subtropical Dry Broadleaf Forests or Tropical and Subtropical Coniferous Forests,
218
10) Temperate Broadleaf and Mixed Forests or Temperate Coniferous Forests. We then
219
rasterized the distribution of these habitats and, for all species with a grid cell of 1°×1°,
220
removed all cells that had more than one type of habitat, counted the number of habitat
221
types found within the range of each species, and classified any species whose range
222
included at least nine of the 10 habitat types as a hyper-generalist. Following these
223
criteria, we removed the following 12 species from the analysis of the distribution by co-
224
occurrence: Sus scrofa (wild boar), Lasiurus cinereus (hoary bat), and 10 carnivores
225
(Canis lupus, Crocuta crocuta, Cuon alpinus, Felis silvestris, Lutra lutra (European
226
otter), Panthera pardus, Puma concolor (cougar), Ursus arctos, Ursus americanus
227
(American black bear) and Vulpes vulpes (red fox)).
228
The change in the distribution of the 185 species whose range was estimated based on co-
229
occurrences with extant species led to an increase from a total of 181 occurrences to
230
117,136 occurrences in 110 × 110 km cells, whereas the change in the distribution of the
231
three species whose range was estimated by co-occurrence with other extinct species led
232
to an increase from a total of 0 occurrences to 2,779 occurrences in 110 × 110 km cells.
233
Fifty species whose ranges were estimated by co-occurrences with extant species had
234
their range expanded by more than 1000 cells. For eight of the species, the expansion was
235
greater than 2000 cells: Arctodus simus (North American short-faced bear), Canis dirus
236
(dire wolf), Coelodonta antiquitatis (woolly rhinoceros), Eremotherium laurillardi (Pan-
237
American ground sloth), Mammuthus primigenius (woolly mammoth), Neochoerus
238
aesopi (a giant capybara), Panthera spelaea (cave lion), and Stephanorhinus
239
kirchbergensis (Merck’s rhinoceros). Two of the species whose ranges were based on co-
240
occurrences with extinct species had their range expanded by more than 1000 cells:
241
Holmesina paulacoutoi (a pampathere) and Tapirus rondoniensis (an extinct South
242
American tapir).
243
Notably, this method will overestimate the distribution of true relictual species that used
244
to co-occur with widespread species, but it may underestimate the distribution of
245
generalist species only known from non-analogous species communities. The combined
246
effect of these two things may potentially balance each other, but researchers using the
247
specific maps may need to be particularly cautious with the maps of Elasmotherium
248
sibiricum (a giant rhinoceros), Equus ovodovi (a horse), Homotherium latidens (European
249
Dirk-toothed cat), Kolpochoerus majus (an African pig), Metridiochoerus compactus (an
250
African pig), Neolicaphrium recens (a litoptern), Pliomys lenki (an European vole),
251
Rusingoryx atopocranion (an African antelope), Soergelia minor (an Eurasian musk-ox),
252
and Tapirus rondoniensis because the estimated ranges are substantially larger than the
253
area geographically encompassing their known late-Quaternary occurrences, and for
254
Doedicurus clavicaudatus (a South American glyptodont) and Tremarctos floridanus
255
(Florida spectacled bear) because their estimated ranges are substantially smaller than
256
similarly expected from their known late-Quaternary occurrences.
257
In order to validate the fossil co-occurrence approach we tried estimating the range of 39
258
North American species within continental United States and Canada based on their fossil
259
occurrences in Faunmap (Graham & Lundelius, 2010). The 39 species were the 11
260
terrestrial species of Artidactyla (Alces americanus (American moose), Antilocapra
261
americana (Pronghorn), Bison bison, Cervus elaphus, Odocoileus hemionus (Mule deer),
262
Odocoileus virginianus (White-tailed deer), Oreamnos americanus (Rocky Mountain
263
goat), Ovibos moschatus, Ovis canadensis (Bighorn sheep), Ovis dalli (Thinhorn sheep)
264
and Rangifer tarandus) and the 28 terrestrial species of Carnivora (Alopex lagopus
265
(Arctic fox), Bassariscus astutus (Ringtail), Canis latrans, Canis lupus, Canis rufus (Red
266
wolf), Gulo gulo, Lontra canadensis (North American river otter), Lynx canadensis
267
(Canadian lynx), Lynx rufus (Bobcat), Martes americana (American marten), Martes
268
pennanti (Fisher), Procyon lotor (Racoon), Mephitis mephitis (Striped skunk), Mustela
269
erminea (Stoat), Mustela frenata (Long-tailed weasel), Mustela nigripes (Black-footed
270
ferret), Mustela nivalis (Least weasel), Neovison vison (American mink), Puma concolor
271
(Cougar), Spilogale gracilis (Western spotted skunk), Spilogale putorius (Eastern spotted
272
skunk), Taxidea taxus (American badger), Urocyon cinereoargenteus (Gray fox), Ursus
273
americanus (American black bear), Ursus arctos (Brown bear), Vulpes macrotis (Kit
274
fox), Vulpes velox (Swift fox) and Vulpes vulpes (Red fox)), which has a substantial part
275
of their current range within continental United States and Canada and fossil occurrences
276
in the Faunmap database. While doing this we used the same taxon specific modifications
277
we used for the extinct species (Supplementary Data 1). For our test case this meant that
278
we did not allow any range overlap between four sets of closely related species (Ovis
279
canadensis / Ovis dalli, Canis lupus / Canis rufus, Spilogale gracilis / Spilogale putorius
280
and Vulpes macrotis / Vulpes velox), that we restricted the occurrence of the members of
281
Caprinae (Oreamnos americanus, Ovis canadensis and Ovis dalli) to mountainous
282
regions, and that we expanded the range of six species (Alces americanus, Bison bison,
283
Canis lupus, Martes americana, Mustela erminea and Mustela nivalis) northwards to
284
include the Labrador peninsula because they have predicted ranges south and north of
285
Laurentide Ice Sheet and on both coasts and the entire peninsula was within their climatic
286
convex hull. A true natural lack from the Labrador peninsula therefore seemed highly
287
unlikely.
288
The diversity we predicted this way was highly correlated with the historic diversity for
289
these species (i.e., the diversity based on their known historical ranges, which we for
290
these species assume to be identical to their natural ranges) (ρ=0.856), and this
291
correlation was higher than the corresponding correlation between the current and
292
historic diversity (ρ=0.762). The average predicted diversity for this validation (total
293
mean 18.6; Artidactyla mean 5.1; carnivora mean 13.5) was also were similar to the
294
values based on their historic diversity (total mean 18.0; Artidactyla mean 4.0; carnivora
295
mean 14.0). Finally to look at the patterns geographically, we compared the overall shape
296
of the diversity gradient between the current, the historic and the predicted species
297
diversity (Figure S11) and found that the co-occurrence predicted diversity pattern
298
accurately represent the placement of the high diversity areas according to the known
299
historical ranges, while the current diversity pattern does not.
300
301
8) Species-specific modifications
302
For 13 species, we opted for unique species-specific solutions because none of the other
303
approaches were appropriate; the data needed to use them could not be found (e.g.,
304
extinct species for which no suitable multi-species fossil site could be found) or we
305
judged that species-specific choices would better capture the natural ranges than the
306
standard method. The species covered here broadly comprise three types: 1) species with
307
very limited knowledge of their historic distribution, but decent knowledge of ecology
308
and general late-Quaternary geography, making it possible to estimate their natural range:
309
Equus ferus (horse) and Lynx pardinus (Iberian lynx); 2) relatively poorly known fossil
310
species with no records from any good multi-species fossil sites: Agalmaceros blicki (a
311
South American deer), Cuvieronius hyodon (a gompthothere), Cryptoprocta spelea (giant
312
fossa), Dactylopsila kambuayai (a New Guinean possum), Elephas iolensis (an extinct
313
African relative of the Asian elephant), and Petauroides ayamaruensis (a New Guinean
314
possum); and 3) poorly known species accepted by IUCN with large uncertainty of the
315
correct location of the type locality Eudorcas rufina (red gazelle), Gerbillus agag (a
316
gerbil), Gerbillus burtoni (a gerbil), Juscelinomys candango (Candango mouse), and
317
Makalata obscura (dusky spiny tree rat). This led to an increase from a total of 11
318
occurrences to 11,132 occurrences in 110 × 110 km cells, largely caused by an increase
319
in Equus ferus from 8 to 9,257 and from 0 to 1,243 for Elephas iolensis.
320
321
322
(B) Esimation of functional diversity
323
For the functional diversity analyses of all species and non-marine species, we focused
324
on the three dimensions of niche space, habitat, body size, and diet, whereas the analysis
325
of the three terrestrial subsets only focused on body size and diet.
326
For habitat, we coded all animals as flying (bats), aquatic (coded as freshwater or marine
327
by IUCN), or terrestrial (non-bats coded as terrestrial by IUCN), with several species
328
coded as both aquatic and terrestrial. Body sizes were initially based on Smith et al.
329
(2005), but supplemented with data from other sources for missing species. In total, we
330
ended with estimated sizes for 5675 species, and the sizes of the remaining 68 species
331
were set as the size of their closest relative with information on body size. All species
332
were categorized into classes of 0.5 log10 units (i.e., one class for species between 100 and
333
100.5 gram and one for species between 100.5 and 101 gram). For diet, we coded all
334
animals as carnivores or herbivores, with omnivores coded as both herbivores and
335
carnivores. Thus, our treatment of functional diversity can be seen as a multidimensional
336
version of the bin-filling approach of (Huang et al., 2012).
337
Diet data were based on Jones et al. (2009), but extrapolated to the entire genus, tribe,
338
subfamily, or family, and occasionally to closely related families if all members with
339
information had the same diet or to members of the same subgenus or species group if the
340
diet appeared constant within this smaller clade but variable in the genus. For genera
341
without information in the database for families without constant diet and for families not
342
listed in the database, we searched for information from elsewhere and included it if
343
possible. Thus, 4552 species could be classified as carnivores (including species eating
344
invertebrates and vertebrates), omnivores, or herbivores. The remaining 1195 species,
345
belonging to genera either without available diet information or with variable diet
346
between species, were given a probability of being in each of the three categories. These
347
probabilities were estimated based on a mean of the means approach, in which each
348
taxonomic group was weighted equally no matter how many species the group contained
349
(which is equivalent to maximum parsimony estimation on a purely taxonomic tree). This
350
is illustrated by a hypothetical subfamily with three genera. Genus A has 10 species, half
351
herbivores and half omnivores. Genus B has one species, an herbivore. The diet of genus
352
C is unknown, but we would assign it a probability of 0.75 for being an herbivore
353
(0.5×prob herbivore (Genus A) + 0.5×prob herbivore (Genus B)).
354
Next, we counted how many classes the species occurring in a given grid cell belonged to
355
based on diet and size (and habitat for all species and non-marine species); one such class
356
could be carnivorous flying mammals (bats) between 100.5 and 101 gram, another could
357
be terrestrial herbivores between 103 and 103.5 gram. The analyses were performed 100
358
times with reassignment of the probability of each diet each time for the 1195 uncertain
359
species and the median value of the analyses assigned to the cell.
360
361
References
362
Arjo, W.M. & Pletscher, D.H. (1999) Behavioral responses of coyotes to wolf
363
recolonization in northwestern Montana. Canadian Journal of Zoology, 77, 1919–1927.
364
Barber, C.B., Habel, K., Grasman, R., Gramacy, R.B., Stahel, A. & Sterratt, D.C. (2013)
365
geometry: Mesh generation and surface tesselation. R package version 0.3-3.
366
http://cran.r-project.org/ (Accessed: 12th February 2015).
367
Bover, P.P. & Alcover, J.A. (2008) Extinction of the autochthonous small mammals of
368
Mallorca (Gymnesic Islands, Western Mediterranean) and its ecological consequences.
369
Journal of Biogeography, 35, 1112-1122.
370
Brummitt, R.K. (2001) World Geographical Scheme for Recording Plant Distributions
371
Edition 2. International Working Group on Taxonomic Databases For Plant Sciences
372
(TDWG) Available at http://www.tdwg.org/ (Accessed: 12th February 2015).
373
Carter, T., Atkinson, M.S. & LaPointe, D.A. (2009) Introduced avian diseases, climate
374
change, and the future of Hawaiian honeycreepers. Journal of Avian Medicine and
375
Surgery, 23, 53-63.
376
Cucchi, T., Vigne, J.D & Auffray, J.C. (2005) First occurrence of the house mouse (Mus
377
musculus domesticus Schwarz & Schwarz, 1943) in the Western Mediterranean: a
378
zooarchaeological revision of subfossil occurrences. Biological Journal of the Linnean
379
Society, 84, 429-445.
380
Ferrusquía-Villafranca, I. (2010) Pleistocene mammals of Mexico: A critical review of
381
regional chronofaunas, climate change response and biogeographic provinciality.
382
Quaternary Intermational, 217, 53–104.
383
Graham, R.W. & Lundelius, E.L. (2010) FAUNMAP II: New data for North America
384
with a temporal extension for the Blancan, Irvingtonian and early Rancholabrean.
385
FAUNMAP II Database, version 1.0. Available at
386
http://www.ucmp.berkeley.edu/faunmap/(Accessed: 12th February 2015).
387
Hall, E.R. (1981) The Mammals of North America. Wiley, New York.
388
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. & Jarvis, A. (2005) Very high
389
resolution interpolated climate surfaces for global land areas. International Journal of
390
Climatology, 25, 1965-197.
391
Jewell, K. (2009) https://stat.ethz.ch/pipermail/r-help/2009-December/415351.html
392
(Accessed: 12th February 2015).
393
Jones, K.E., et al. (2009) PanTHERIA: a species-level database of life history, ecology,
394
and geography of extant and recently extinct mammals. Ecology, 90, 184.
395
Justina, R., Hunter, L. & Zigouris, J. (2005) Setting conservation and research priorities
396
for larger African carnivores. WCS Working Paper 24. Wildlife Conservation Society,
397
New York.
398
Mosbrugger, V. & Utescher, T. (1997) The coexistence approach - a method for
399
quantitative reconstructions of Tertiary terrestrial palaeoclimate data using plant fossils.
400
Palaeogeography, Palaeoclimatology, Palaeoecology, 134, 61-86.
401
Olson, D.M. (2001) Terrestrial ecoregions of the world: A new map of life on Earth.
402
BioScience, 51, 933-938.
403
Paleobiology Database (2015) Available at http://paleodb.org/ (Accessed: 12th February
404
2015).
405
Smith, F.A., Lyons, S.K., Ernest, S.K.M., Jones, K.E., Kaufman, D.M., Dayan, T.,
406
Marquet, P.A., Brown, J.H. & Haskell, J.P. (2005) Body mass of late Quaternary
407
mammals. Ecology, 84,3403.
408
Stringer, C. (2012) Evolution: What makes a modern human. Nature, 485, 33-35.
409
Supplementary figures
410
Figure S1
411
Estimated current and present natural diversities for all mammals.
412
413
Figure S2
414
Estimated current and present natural diversities for all non-marine mammals.
415
416
Figure S3
417
Estimated current and present natural diversities for all terrestrial mammals.
418
419
Figure S4
420
421
Estimated current and present natural diversities for all large terrestrial mammals (>10
kg).
422
423
Figure S5
424
Estimated current and present natural diversities for all terrestrial megafauna (>44.5 kg).
425
426
Figure S6
427
Estimated total, historic and prehistoric loss for all mammals.
428
429
Figure S7
430
Estimated total, historic and prehistoric loss for all non-marine mammals.
431
432
Figure S8
433
Estimated total, historic and prehistoric loss for all terrestrial mammals.
434
435
Figure S9
436
Estimated total, historic and prehistoric loss for all large terrestrial mammals (>10 kg).
437
438
Figure S10
439
Estimated total, historic and prehistoric loss for all terrestrial megafauna (>44.5 kg).
440
441
Figure S11
442
443
The current diversity, the historic diversity and the predicted diversity based on the fossil
co-occurrence approach for 39 North American mammals.
444
445
Figure S12
446
The relationship between species diversity and NDVI.
447
448
Figure S13
449
The relationship between phylogenetic diversity and NDVI.
450
451
Figure S14
452
The relationship between functional diversity and NDVI.
453
454
Figure S15
455
The relationship between species diversity and elevational range.
456
457
Figure S16
458
The relationship between phylogenetic diversity and elevational range.
459
460
Figure S17
461
The relationship between functional diversity and elevational range.
Table S1. The relationship between diversity and environmental drivers for species diversity for the five different species sets. Standardized
parameter estimates for a full SAR model are given along with standard errors of the estimate in parentheses. All P-values are based on Wald’s
tests
Terrestrial megafauna
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
-0.031 (0.080)NS
-0.061 (0.083)NS
-0.062 (0.095)NS
NS
***
Elevational range
0.019 (0.016)
0.067 (0.017)
0.102 (0.018)***
Temperature
0.056 (0.069)NS
0.107 (0.073)NS
-0.041 (0.079)NS
log(precipitation)
0.002 (0.044)NS
0.070 (0.046)NS
0.039 (0.049)NS
Precipitation seasonality
-0.014 (0.022)NS
0.047 (0.023)*
0.021 (0.025)NS
Temperature seasonality
0.097 (0.062)NS
0.071 (0.066)NS
-0.066 (0.071)NS
NDVI
0.509 (0.013)***
0.427 (0.014)***
0.327 (0.015)***
Open areas
0.076 (0.028)*
0.157 (0.030)***
0.114 (0.032)***
log(precipitation): Open Areas
0.130 (0.041)**
0.104 (0.044)**
0.122 (0.047)**
Temperature: Open Areas
0.027 (0.030)NS
0.086 (0.032)**
0.103 (0.035)**
Pseudo R2
0.570
0.425
0.402
Large terrestrial species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
-0.023 (0.068)NS
-0.047 (0.076)NS
-0.058 (0.084)NS
Elevational range
0.028 (0.015)NS
0.062 (0.015)***
0.087 (0.017)***
Temperature
0.102 (0.065)NS
0.133 (0.068)*
0.027 (0.073)NS
log(precipitation)
0.012 (0.042)NS
0.065 (0.043)NS
0.063 (0.047)NS
Precipitation seasonality
-0.018 (0.021)NS
0.017 (0.022)NS
-0.000 (0.024)NS
Temperature seasonality
0.069 (0.059)NS
0.059 (0.061)NS
-0.041 (0.066)NS
NDVI
0.547 (0.013)***
0.451 (0.013)***
0.383 (0.014)***
Open areas
0.098 (0.028)***
0.153 (0.028)***
0.128 (0.030)***
log(precipitation): Open Areas
0.166 (0.040)***
0.134 (0.041)**
0.112 (0.044)*
Temperature: Open Areas
0.040 (0.030)NS
0.084 (0.030)**
0.108 (0.033)***
Pseudo R2
0.666
0.483
0.466
All terrestrial species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
-0.014 (0.053)NS
-0.024 (0.055)NS
-0.019 (0.058)NS
Elevational range
0.082 (0.014)***
0.104 (0.014)***
0.114 (0.015)***
Temperature
0.108 (0.058)NS
0.122 (0.061)*
0.074 (0.064)NS
log(precipitation)
0.104 (0.040)**
0.136 (0.041)***
0.150 (0.043)***
Precipitation seasonality
-0.025 (0.020)NS
-0.010 (0.020)NS
-0.018 (0.021)NS
Temperature seasonality
-0.019 (0.052)NS
-0.032 (0.054)NS
-0.068 (0.056)NS
NDVI
0.666 (0.013)***
0.659 (0.013)***
0.616 (0.014)***
Open areas
0.040 (0.026)NS
0.052 (0.027)NS
0.041 (0.028)NS
log(precipitation): Open Areas
0.048 (0.038)NS
0.018 (0.039)NS
0.003 (0.041)NS
Temperature: Open Areas
0.009 (0.028)NS
0.021 (0.029)NS
0.038 (0.030)NS
Pseudo R2
0.775
0.740
0.734
Non-marine species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
0.003 (0.050)NS
0.001 (0.051)NS
0.004 (0.054)NS
Elevational range
0.064 (0.013)***
0.076 (0.013)***
0.082 (0.014)***
Temperature
0.102 (0.056)NS
0.112 (0.057)NS
0.086 (0.059)NS
log(precipitation)
0.081 (0.036)*
0.095 (0.038)*
0.096 (0.039)*
Precipitation seasonality
-0.039 (0.019)*
-0.031 (0.020)NS
-0.036 (0.020)NS
Temperature seasonality
-0.182 (0.051)***
-0.214 (0.052)***
-0.235 (0.054)***
NDVI
0.658 (0.013)***
0.654 (0.013)***
0.621 (0.014)***
Open areas
-0.009 (0.025)NS
-0.007 (0.026)NS
-0.016 (0.027)NS
log(precipitation): Open Areas
0.057 (0.036)NS
0.044 (0.037)NS
0.039 (0.038)NS
Temperature: Open Areas
-0.026 (0.027)NS
-0.026 (0.028)NS
-0.018 (0.029)NS
Pseudo R2
0.763
0.750
0.749
All species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
0.008 (0.052)NS
0.007 (0.052)NS
0.010 (0.055)NS
***
***
Elevational range
0.088 (0.013)
0.102 (0.014)
0.108 (0.014)***
Temperature
0.150 (0.057)**
0.161 (0.059)**
0.136 (0.061)*
log(precipitation)
0.110 (0.034)**
0.129 (0.036)***
0.129 (0.037)***
Precipitation seasonality
-0.018 (0.020)NS
-0.009 (0.021)NS
-0.014 (0.022)NS
Temperature seasonality
-0.228 (0.052)***
-0.268 (0.054)***
-0.289 (0.056)***
NDVI
0.593 (0.015)***
0.575 (0.016)***
0.532 (0.016)***
Open areas
-0.027 (0.026)NS
-0.028 (0.028)NS
-0.037 (0.028)NS
log(precipitation): Open Areas
0.039 (0.035)NS
0.025 (0.037)NS
0.019 (0.037NS
Temperature: Open Areas
-0.029 (0.028)NS
-0.029 (0.030)NS
-0.018 (0.030)NS
Pseudo R2
0.717
0.704
0.699
NS
P > 0.05; * 0.05 > P>0.01; **0.01 > P > 0.001; *** P<0.001
Table S2. The relationship between diversity and environmental drivers for phylogenetic diversity for the five different species sets.
Standardized parameter estimates for a full SAR model are given along with standard errors of the estimate in parentheses. All P-values are
based on Wald’s tests
Terrestrial megafauna
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
-0.099 (0.077)NS
-0.096 (0.097)NS
0.027 (0.108)NS
NS
NS
Elevational range
0.014 (0.016)
0.025 (0.018)
0.095 (0.019)***
Temperature
0.168 (0.065)**
0.223 (0.073)**
0.063 (0.078)NS
log(precipitation)
-0.028 (0.044)NS
0.066 (0.048)NS
0.033 (0.051)NS
Precipitation seasonality
-0.019 (0.021)NS
0.009 (0.023)NS
-0.030 (0.025)NS
Temperature seasonality
0.130 (0.058)*
0.054 (0.064)NS
-0.068 (0.069)NS
NDVI
0.632 (0.013)***
0.506 (0.014)***
0.417 (0.015)***
Open areas
0.052 (0.030)NS
0.136 (0.033)***
0.009 (0.035)NS
log(precipitation): Open Areas
0.144 (0.043)***
0.131 (0.047)**
0.178 (0.050)***
Temperature: Open Areas
-0.033 (0.031)NS
0.035 (0.034)NS
0.044 (0.036)NS
Pseudo R2
0.692
0.487
0.483
Large terrestrial species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
-0.019 (0.058)NS
-0.041 (0.070)NS
-0.038 (0.075)NS
Elevational range
0.031 (0.015)*
0.060 (0.016)***
0.086 (0.016)***
Temperature
0.135 (0.062)*
0.216 (0.068)**
0.140 (0.069)*
log(precipitation)
0.024 (0.042)NS
0.068 (0.045)NS
0.089 (0.045)*
Precipitation seasonality
-0.031 (0.021)NS
-0.009 (0.022)NS
-0.017 (0.023)NS
Temperature seasonality
0.103 (0.057)NS
0.105 (0.062)NS
0.007 (0.064)NS
NDVI
0.645 (0.012)***
0.558 (0.013)***
0.476 (0.013)***
Open areas
0.072 (0.028)**
0.127 (0.029)***
0.093 (0.029)**
log(precipitation): Open Areas
0.103 (0.040)**
0.099 (0.042)*
0.087 (0.043)*
Temperature: Open Areas
0.007 (0.029)NS
0.054 (0.031)NS
0.075 (0.031)*
Pseudo R2
0.697
0.533
0.511
All terrestrial species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
-0.064 (0.056)NS
-0.064 (0.060)NS
-0.044 (0.064)NS
Elevational range
0.061 (0.012)***
0.077 (0.013)***
0.086 (0.013)***
Temperature
0.247 (0.052)***
0.290 (0.054)***
0.251 (0.056)***
log(precipitation)
0.048 (0.035)NS
0.068 (0.036)NS
0.076 (0.037)*
Precipitation seasonality
-0.022 (0.017)NS
-0.016 (0.017)NS
-0.018 (0.018)NS
Temperature seasonality
0.105 (0.045)*
0.068 (0.047)NS
0.039 (0.049)NS
NDVI
0.740 (0.011)***
0.742 (0.012)***
0.712 (0.012)***
Open areas
0.056 (0.024)*
0.079 (0.025)**
0.053 (0.025)*
log(precipitation): Open Areas
0.088 (0.034)*
0.084 (0.036)*
0.078 (0.037)*
Temperature: Open Areas
-0.021 (0.026)NS
-0.014 (0.027)NS
-0.001 (0.027)NS
Pseudo R2
0.828
0.812
0.789
Non-marine species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
-0.041 (0.052)NS
-0.039 (0.054)NS
-0.025 (0.057)NS
Elevational range
0.059 (0.012)***
0.068 (0.012)***
0.074 (0.012)***
Temperature
0.273 (0.051)***
0.301 (0.053)***
0.278 (0.054)***
log(precipitation)
0.032 (0.034)NS
0.044 (0.035)NS
0.046 (0.036)NS
Precipitation seasonality
-0.024 (0.016)NS
-0.018 (0.017)NS
-0.021 (0.017)NS
Temperature seasonality
-0.031 (0.045)NS
-0.076 (0.046)NS
-0.099 (0.048)*
NDVI
0.747 (0.012)***
0.742 (0.012)***
0.716 (0.012)***
Open areas
0.033 (0.023)NS
0.045 (0.024)NS
0.026 (0.025)NS
log(precipitation): Open Areas
0.095 (0.033)**
0.092 (0.034)**
0.091 (0.035)*
Temperature: Open Areas
-0.046 (0.025)NS
-0.044 (0.026)NS
-0.036 (0.026)NS
Pseudo R2
0.827
0.816
0.794
All species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
-0.014 (0.046)NS
-0.009 (0.048)NS
0.012 (0.052)NS
***
***
Elevational range
0.055 (0.012)
0.067 (0.012)
0.072 (0.013)***
Temperature
0.190 (0.048)***
0.220 (0.051)***
0.204 (0.052)***
log(precipitation)
0.092 (0.029)**
0.109 (0.031)***
0.110 (0.032)***
Precipitation seasonality
0.002 (0.017)NS
0.013 (0.017)NS
0.009 (0.018)NS
Temperature seasonality
-0.084 (0.043)NS
-0.136 (0.046)**
-0.156 (0.047)**
NDVI
0.757 (0.013)***
0.750 (0.014)***
0.727 (0.014)***
Open areas
0.027 (0.023)NS
0.038 (0.024)NS
0.022 (0.025)NS
log(precipitation): Open Areas
0.060 (0.030)NS
0.056 (0.032)NS
0.051 (0.033)NS
Temperature: Open Areas
-0.026 (0.024)NS
-0.026 (0.026)NS
-0.019 (0.026)NS
Pseudo R2
0.819
0.801
0.780
NS P > 0.05; * 0.05 > P>0.01; **0.01 > P > 0.001; *** P<0.001
Table S3. The relationship between diversity and environmental drivers for functional diversity for the five different species sets. Standardized
parameter estimates for a full SAR model are given along with standard errors of the estimate in parentheses. All P-values are based on Wald’s
tests
Terrestrial megafauna
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
0.004 (0.049)NS
-0.019 (0.062)NS
0.020 (0.069)NS
NS
NS
Elevational range
0.021 (0.015)
0.033 (0.017)
0.102 (0.019)***
Temperature
0.091 (0.061)NS
0.208 (0.071)**
0.096 (0.080)NS
log(precipitation)
0.017 (0.043)NS
0.032 (0.048)NS
0.073 (0.055)NS
Precipitation seasonality
0.007 (0.021)NS
0.031 (0.024)NS
0.018 (0.027)NS
Temperature seasonality
0.276 (0.054)***
0.108 (0.063)NS
-0.031 (0.071)NS
NDVI
0.688 (0.013)***
0.547 (0.014)***
0.442 (0.017)***
Open areas
0.051 (0.028)NS
0.078 (0.032)*
0.039 (0.036)NS
log(precipitation): Open Areas
0.104 (0.041)*
0.121 (0.046)**
0.106 (0.052)*
Temperature: Open Areas
0.022 (0.030)NS
0.009 (0.033)NS
-0.002 (0.038)NS
Pseudo R2
0.681
0.480
0.462
Large terrestrial species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
0.002 (0.045)NS
-0.018 (0.054)NS
0.031 (0.057)NS
Elevational range
0.018 (0.014)NS
0.028 (0.015)NS
0.062 (0.017)***
Temperature
0.123 (0.058)*
0.203 (0.064)**
0.091 (0.070)NS
log(precipitation)
-0.005 (0.042)NS
0.006 (0.044)NS
0.079 (0.048)NS
Precipitation seasonality
0.004 (0.020)NS
0.043 (0.022)*
0.036 (0.024)NS
Temperature seasonality
0.301 (0.052)***
0.275 (0.057)***
0.287 (0.062)***
NDVI
0.713 (0.013)***
0.615 (0.014)***
0.538 (0.015)***
Open areas
0.030 (0.028)NS
0.050 (0.029)NS
0.043 (0.032)NS
log(precipitation): Open Areas
0.110 (0.040)**
0.108 (0.042)*
0.095 (0.047)*
Temperature: Open Areas
-0.006 (0.029)NS
-0.004 (0.031)NS
0.026 (0.034)NS
Pseudo R2
0.722
0.575
0.558
All terrestrial species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
-0.019 (0.036)NS
-0.022 (0.033)NS
-0.015 (0.034)NS
Elevational range
0.070 (0.012)***
0.080 (0.013)***
0.089 (0.012)***
Temperature
0.227 (0.049)***
0.267 (0.049)***
0.211 (0.050)***
log(precipitation)
0.036 (0.037)NS
0.036 (0.037)NS
0.072 (0.037)*
Precipitation seasonality
0.022 (0.018)NS
0.035 (0.017)*
0.031 (0.017)NS
Temperature seasonality
0.390 (0.044)***
0.396 (0.043)***
0.399 (0.043)***
NDVI
0.812 (0.012)***
0.794 (0.012)***
0.768 (0.012)***
Open areas
0.059 (0.025)*
0.068 (0.025)**
0.050 (0.025)*
log(precipitation): Open Areas
0.059 (0.036)NS
0.081 (0.036)*
0.051 (0.036)NS
Temperature: Open Areas
0.013 (0.027)NS
0.007 (0.027)NS
0.033 (0.027)NS
Pseudo R2
0.849
0.829
0.820
Non-marine species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
0.031 (0.042)NS
0.011 (0.030)NS
0.016 (0.032)NS
Elevational range
0.034 (0.011)**
0.037 (0.011)***
0.046 (0.011)***
Temperature
0.169 (0.046)***
0.179 (0.045)***
0.146 (0.048)**
log(precipitation)
0.040 (0.030)NS
0.047 (0.031)NS
0.066 (0.032)*
Precipitation seasonality
-0.003 (0.015)NS
-0.002 (0.016)NS
-0.007 (0.016)NS
Temperature seasonality
0.166 (0.041)***
0.134 (0.040)***
0.138 (0.041)***
NDVI
0.811 (0.011)***
0.807 (0.011)***
0.787 (0.012)***
Open areas
0.020 (0.022)NS
0.031 (0.022)NS
0.014 (0.023)NS
log(precipitation): Open Areas
0.100 (0.031)**
0.116 (0.032)***
0.093 (0.032)**
Temperature: Open Areas
0.007 (0.023)NS
-0.003 (0.024)NS
0.013 (0.024)NS
Pseudo R2
0.861
0.854
0.839
All species
Total natural diversity
Natural IUCN diversity
Current diversity
Intercept
0.019 (0.054)NS
-0.011 (0.039)NS
-0.009 (0.041)NS
**
**
Elevational range
0.049 (0.015)
0.047 (0.016)
0.060 (0.016)***
Temperature
0.050 (0.061)NS
-0.047 (0.061)NS
-0.087 (0.063)NS
log(precipitation)
0.117 (0.038)**
0.133 (0.041)**
0.157 (0.042)***
Precipitation seasonality
0.019 (0.021)NS
0.024 (0.023)NS
0.012 (0.023)NS
Temperature seasonality
0.011 (0.054)NS
-0.042 (0.054)NS
-0.042 (0.056)NS
NDVI
0.750 (0.017)***
0.748 (0.018)***
0.720 (0.018)***
Open areas
0.022 (0.030)NS
0.050 (0.032)NS
0.035 (0.033)NS
log(precipitation): Open Areas
0.059 (0.039)NS
0.084 (0.043)NS
0.054 (0.044)NS
Temperature: Open Areas
0.022 (0.032)NS
0.023 (0.034)NS
0.040 (0.035)NS
Pseudo R2
0.724
0.708
0.682
NS P > 0.05; * 0.05 > P>0.01; **0.01 > P > 0.001; *** P<0.001
462
Download