SUPPLEMENTARY INFORMATION
Species diversity of bats along an altitudinal gradient of Mount Mulanje, southern Malawi Michael Curran, Mirjam Kopp, Jan Beck and Jakob Fahr Please cite the original study when using any of the data contained herein.
EXPANDED METHODS Echolocation reference calls Echolocation reference calls were recorded with a D240X time-expansion bat detector (Pettersson Electronics, Sweden) connected to a Cowon iAudio digital recording device, which were subsequently analysed with the software Raven Pro (Charif, R. A., Waak, A. M. and Strickman, L. M. 2008. Cornell Laboratory of Ornithology). Reference calls were taken from a representative sample of all echolocating species (M. Kopp, unpublished data), but were only used in this study to aid identification of Miniopterus species. A combination of forearm measurements, minimum and peak frequency of echolocation calls, and genetic sequence data (only for Miniopterus sp. 2–4; B. Appleton unpubl. data) using mitochondrial genes ND2 and cytochrome b) was used to differentiate between the species (data not shown). Additional explanatory data We derived a series of additional environmental datasets that have been postulated to be important in determining altitude-richness patterns. Mean annual temperature and relative humidity values were interpolated for each site from nearby weather stations using as automated procedure with the commercial software MeteoNorm (see www.meteonorm.com for details and data sources). Each site's environmental characteristics (longitude, latitude, altitude, position within the gorge and aspect) were used as input variables for the interpolation. We also measured temperature and relative humidity directly in the field during the sampling period using a standard mercury thermometer and humidity gauge, and calculated the mean nightly temperature between the times of 18:00 and 24:00 (4 – 6 measurements per night, mean = 23.14 measurements per site, s.d. = 2.85). In order to assess indirect area effects driven by the species-area relationship (see Romdal & Grytness 2007) we derived two area-
based datasets. Using a GIS (Geographic Resource Analysis Support System, ver. 6.4.1; grass.fbk.eu/), we calculated the area within altitudinal bands of 100 m centred around each site using the Shuttle Radar Topography Mission (SRTM) 90 m resolution Digital Elevation Model (www2.jpl.nasa.gov/srtm/). For the bottom band (580 – 680 m), we clipped the region extent at roughly 100 km around the massif. We also calculated forest extent within each altitudinal band as represented by categories 40 through 100 of the ESA Globecover 2009 landcover dataset (www.esa.int). This step was conducted to better approximate habitat area for forest species. As a proxy for productivity, we derived a Normalized Difference Vegetation Index (Hurlbert & Haskell 2003) as an average of 3 × 3 pixels using bands 3 and 4 from orthorectified Landsat ETM+ images (available at www.landsat.org), acquired on the 9 July 2002. Most predictor data were log-transformed to conform to a normal distribution, and standardized to facilitate comparison of slope values across predictors (species richness was also log-transformed when regressed against area variables). Two variables still deviated significantly from a normal distribution after log-transformation: area of 100 m altitudinal bands (Shapiro normality test, P = 0.009), and forest area within 100 m altitudinal bands (Shapiro normality test, P = 0.006). In both cases, observation of residual plots showed deviation was driven by the lowest area band, but we retained these data because deviation was not extreme (i.e. Cook's distance was less than 1 in a residual vs. leverage plot of the model). We then regressed these variables against observed and estimated species richness (which were logtransformed for regressions against area and temperature) and Fishers α. The table also includes the vegetation indices derived in this study for comparative purposes, and two proxies of sampling intensity (number of samples per site, and number of individuals per site) in order to assess the degree of possible sampling bias associated with the various indicators of diversity.
NEW RECORDS Sixty-three bat species were previously known to occur in Malawi (Bergmans & van Strien 2004, Happold et al. 1988, Happold & Happold 1997, Kock et al. 1998, MillerButterworth et al. 2005, Šklíba et al. 2007). Our study added five new species to the country checklist: Myonycteris relicta, Mops cf. brachypterus, Tadarida aegyptiaca, Mimetillus moloneyi, and Miniopterus cf. minor, to give a total of 68 species. The last of these, M. minor, should be treated with caution as its identity has not yet been confirmed using molecular techniques (as was the case for the other three species of Miniopterus).
ADDITIONAL TABLES AND FIGURES S1. External field measurements for 31 bat species recorded during the study. The statistics include data from opportunistic sampling at additional sites on Mount Mulanje, and also data from nearby mountains in Northern Mozambique (published in Monadjem et al. 2010). s.d. = standard deviation, n = sample size, min, max = range.
Species Pteropodidae Epomophorus crypturus Epomophorus wahlbergi Lissonycteris angolensis Myonycteris relicta Rousettus aegyptiacus Eidolon helvum Rhinolophidae Rhinolophus blasii Rhinolophus clivosus Rhinolophus fumigatus Rhinolophus hildebrandtii Rhinolophus simulator Hipposideridae Hipposideros ruber Vespertilionidae Kerivoula argentata Kerivoula lanosa Myotis tricolor Myotis welwitschii Eptesicus hottentotus Laephotis botswanae Mimetillus moloneyi Neoromicia nana Pipistrellus hesperidus Pipistrellus grandidieri Scotophilus dinganii Miniopteridae Miniopterus sp Miniopterus sp2 Miniopterus sp3 Miniopterus sp4 Molossidae Mops cf. brachypterus Tadarida aegyptiaca Nycteridae Nycteris hispida Nycteris thebaica
Forearm length (mm) mean s.d. n min max
body mass (g) mean s.d. n
min
max
tibia length (mm) mean s.d. n min max
82.1 83.2 84.4 70.4 91.9 123.3
2.8 5 2.4 32 2.1 12 1 4.7 66 3.2 2
79.7 78.9 81.0 70.4 81.9 121.0
85.1 88.8 87.9 70.4 99.8 125.5
84.2 104.0 102.1 65.0 117.9 313.3
15.0 5 10.4 29 12.8 11 1 18.0 65 18.0 2
71.0 77.5 90.0 65.0 88.0 300.5
102.0 121.0 127.5 65.0 168.0 326.0
32.7 34.4 34.4 27.8 39.7 51.4
2.5 3 1.8 17 0.8 10 1 4.8 17 2.3 2
30.3 31.1 33.0 27.8 30.5 49.8
35.2 38.6 35.4 27.8 47.5 53.0
45.5 52.9 50.4 65.4 44.4
0.9 1.2 0.8 1.2 1.2
41.0 49.5 49.3 64.0 42.4
48.0 56.0 51.0 66.7 46.7
9.5 14.3 17.0 31.1 8.4
1.1 1.8 2.7 4.2 1.2
347 76 4 4 29
7.0 12.0 13.0 27.5 7.0
13.0 23.0 19.0 37.0 10.5
18.6 22.2 21.7 27.5 18.3
0.7 0.7 0.5 5.0 1.2
17.3 20.9 21.2 20.1 16.0
20.6 24.0 22.1 30.5 20.7
52.4
0.8 5
51.3
53.3
12.5
2.3
5
9.0
14.5
22.6
1.0 5
21.1 24.0
36.4 31.5 49.7 58.9 47.9 37.0 31.5 31.2 32.0 36.2 52.6
1 1.1 7 1.2 31 1 1.4 2 0.9 17 1 0.9 16 0.9 23 1.6 2 0.5 4
36.4 29.5 47.6 58.9 46.9 35.1 31.5 29.4 30.1 35.0 51.8
36.4 33.1 53.6 58.9 48.9 38.2 31.5 32.3 33.3 37.3 53.0
8.0 4.9 12.9 14.5 13.8 7.6 12.0 3.7 5.7 10.0 21.2
0.7 1.5 0.7 6.0
1 7 30 1 2 17 1 16 23 2 3
8.0 4.0 10.0 14.5 13.5 6.0 12.0 3.0 3.0 9.5 17.0
8.0 7.0 16.0 14.5 14.0 9.0 12.0 5.5 9.0 10.5 28.0
16.6 13.0 22.8 24.6 19.2 15.0 11.6 13.4 12.7 15.1 21.4
1 0.7 3 0.8 27 1 0.2 2 0.6 13 1 0.5 13 0.6 18 0.8 2 1.1 4
16.6 12.3 21.2 24.6 19.0 13.9 11.6 12.6 11.6 14.5 19.9
33.7 43.7 44.0 46.3
2.6 0.9 0.5 0.7
31.8 42.0 43.1 44.3
37.4 46.1 44.4 48.4
6.0 9.6 10.4 14.1
2.2 1.2 3.2 1.2
3 29 5 41
4.5 7.5 8.5 12.0
8.5 11.0 16.0 17.5
13.7 18.4 18.7 20.6
1 13.7 13.7 0.7 27 17.4 19.4 0.6 4 18.0 19.4 0.6 31 19.4 23.4
36.1 48.6
1 0.7 2
36.1 48.1
36.1 49.1
18.0 20.3
4.6
1 2
18.0 17.0
18.0 23.5
15.3
0.1 2
15.2 15.4
39.2 46.0
1.4 2 1.7 8
38.2 42.9
40.2 48.5
7.5 10.9
0.7 1.3
2 7
7.0 9.2
8.0 13.5
19.2 24.9
0.1 2 0.7 6
19.1 19.2 24.0 25.7
351 79 4 4 30
4 32 5 41
1.0 1.4 0.4 0.7
30 40 3 4 11
16.6 13.7 25.0 24.6 19.3 15.9 11.6 14.3 14.2 15.7 22.4
S2. Diversity indices, with 95% confidence interval (CI) for De. Alt
Canopy data excluded
Canopy data included
630
Fisher’s α 4.41
De 7.81
De 95% CI 4.51–11.10
Fisher’s α 4.69
De 8.77
De 95% CI 5.52–12.03
720
6.11
12.12
7.66–16.57
5.93
11.65
7.02–16.28
900
1.94
3.10
1.33–4.87
2.94
4.77
2.47–7.06
1030
2.94
2.50
1.44–3.57
3.73
3.43
2.03–4.83
1220
3.11
3.87
2.06–5.68
4.26
4.85
2.63–7.07
1320
1.88
3.96
2.65–5.26
3.49
6.54
4.25–8.82
1850
0.75
1.69
0.34–3.04
n.a.
n.a.
n.a.
2010
2.02
3.90
1.24–6.56
1.97
3.93
1.75–6.11
S3. Site-specific, non-parametric species richness estimates. Strike-through values were excluded from the calculation of mean and median because they fall below Sobs. F1, F2, F3 refer to singletons, doubletons and tripletons respectively. Q1, Q2, Q3 refer to uniques, duplicates and triplicates respectively. Letters “I” and “A” following Jackknife estimators indicate incidence- or abundance-based variants, respectively. a) All captures Data properties
Estimator values
Coverage (%) Choice
Altitude
n
Sobs
N
F1 F2 F3 Q1 Q2 Q3
ACE ICE Chao1 Chao2 Jack1I Jack1A Jack2I Jack2A Jack3I Jack3A Boot MM
Mean Median
630
6
12
56
4
16.8 15.5 13.5
720
6
18
140
4
3
1
7
6
3
19.6 19.6 18.8
20.5
900
5
10
85
4
2
0
4
2
1
16.0 14.0 12.0
11.6
1030
5
16
268
5
2
4
6
4
4
19.4 21.7 19.3
18.4
1220
6
14
110
6
2
3
10 3
0
19.7 43.7 19.0
1320
8
8
31
2
2
2
4
1
1
9.3
13.0 8.3
2010
11 5
23
2
1
0
2
1
0
8.2
7.1
3
0
4
5
0
7.0
12.8
15.3
Sest
15.3
15.3
16.4
13.8
16.3
13.8 17.5
78.9
78.3
Jack1A 15.3
23.8
20.3
25.3
21.4
23.8
21.4
19.6 29.5
82.0
86.0
Jack1A 20.3
13.2
13.2
14.7
14.7
15.4
15.3
11.5 13.4
72.8
73.1
Jack2A 14.7
20.8
19.0
20.7
21.1
22.3
22.4
18.3 22.4
78.1
77.2
Jack1A 19.0
30.7
22.3
18.0
27.4
20.9
29.3
22.8
17.6 34.8
54.8
62.1
Jack3A 22.8
16.0
11.5
9.8
13.9
10.0
15.4
9.4
9.5
12.3
69.4
74.5
Jack2A 10.0
7.0
6.8
6.8
7.7
7.7
7.9
7.9
5.8
6.9
69.0
71.1
Jack2A 7.7
c) Animalivorous species only Data properties
Estimator values
Coverage (%) Choice
Altitude
n
Sobs
N
F1 F2 F3 Q1 Q2 Q3
ACE ICE Chao1 Chao2 Jack1I Jack1A Jack2I Jack2A Jack3I Jack3A Boot MM
Mean Median
630
6
10
44
4
3
0
4
3
0
14.7 14.2 11.5
74.4
11.3
13.3
13.3
14.4
14.4
14.3
14.3
11.6 14.0
71.0
Sest
Jack1A 13.3
720
6
14
75
2
4
0
6
5
2
15.0 19.0 14.2
16.1
19.0
15.7
20.3
14.9
20.0
13.1
16.5 25.6
78.5
85.0
Jack2A 14.9
900
5
7
70
3
2
0
3
2
1
10.8 10.5 8.0
9.3
9.4
9.4
10.3
10.3
10.6
10.5
8.1
71.9
69.9
Jack2A 10.3
1030
5
11
237
2
2
3
3
3
2
11.9 13.2 11.3
11.6
13.4
12.6
13.9
12.9
13.7
12.9
12.3 14.5
85.6
85.3
Jack1A 12.6
1220
6
9
100
4
1
2
5
3
0
13.2 17.9 12.0
13.2
13.2
12.3
14.9
14.5
15.3
15.9
10.9 17.1
63.4
65.1
Jack3A 14.5
1320
8
5
26
0
2
1
1
1
1
5.0
5.5
5.0
5.0
5.9
5.0
6.0
3.7
5.7
1.8
5.5
6.7
98.8
95.5
MM
2010
11 5
23
2
1
0
2
1
0
8.2
7.0
7.0
5.5
6.8
6.8
7.7
7.7
7.9
7.9
5.8
6.9
70.3
71.5
Jack1A 6.8
9.7
6.7
S4. Assessment of estimator precision by plotting estimator convergence as samples accumulate. Stability across the final two samples was assumed to indicate a precise value.
S5. Habitat structure was estimated based on the frequency of discreet variable classes (vegetation density and canopy cover) occurring at each sample site (a). These were standardized by expressing each site score in terms of standard deviations from the gradient mean (b), and used to construct a distance matrix between all sites (c). a) Percentage scores for each habitat class per site (number of data points) Altitude Veg1
Veg2
Veg3
Can0
Can1
Can2
Can3
Can4
Can5
630
3.39
69.49
27.12
3.33
10.00
16.67
23.33
43.33
3.33
720
0.00
41.18
58.82
0.00
6.67
20.00
40.00
33.33
0.00
900
13.33
28.33
58.33
6.67
10.00
16.67
20.00
33.33
13.33
1030
6.25
43.75
50.00
0.00
8.33
4.17
20.83
37.50
29.17
1220
0.00
30.56
69.44
0.00
5.56
16.67
27.78
22.22
27.78
1320
0.00
66.67
33.33
14.29
28.57
0.00
7.14
21.43
28.57
1850
13.56
62.71
23.73
3.57
14.29
10.71
28.57
39.29
3.57
2010
28.57
26.19
45.24
42.86
14.29
19.05
9.52
14.29
0.00
b) Normalized habitat scores per site Altitude Veg1
Veg2
Veg3
Can0
Can1
Can2
Can3
Can4
Can5
630
-0.48
1.31
-1.13
-0.38
-0.30
0.50
0.11
1.26
-0.74
720
-0.81
-0.28
0.79
-0.61
-0.76
0.95
1.69
0.27
-0.99
900
0.52
-1.00
0.76
-0.15
-0.30
0.50
-0.20
0.27
0.01
1030
-0.19
-0.13
0.26
-0.61
-0.53
-1.20
-0.12
0.68
1.20
1220
-0.81
-0.87
1.44
-0.61
-0.91
0.50
0.53
-0.82
1.09
1320
-0.81
1.15
-0.75
0.37
2.23
-1.77
-1.42
-0.90
1.15
1850
0.54
0.93
-1.34
-0.36
0.28
-0.31
0.61
0.86
-0.72
2010
2.05
-1.12
-0.03
2.34
0.28
0.82
-1.19
-1.61
-0.99
c) Euclidean distance matrix for habitat structure based on standardized scores Altitude 630
720
900
1030
1220
1320
1850
2010
630
–
3.21
3.41
3.35
4.45
4.80
1.62
5.59
720
3.21
–
2.74
3.69
2.81
6.11
3.49
5.58
900
3.41
2.74
–
2.50
2.40
4.97
3.27
3.89
1030
3.35
3.69
2.50
–
2.84
4.02
3.16
5.54
1220
4.45
2.81
2.40
2.84
–
5.36
4.59
5.34
1320
4.80
6.11
4.97
4.02
5.36
–
4.41
5.78
1850
1.62
3.49
3.27
3.16
4.59
4.41
–
5.11
2010
5.59
5.58
3.89
5.54
5.34
5.78
5.11
–
S6. Capture success (number of captures per standardized sampling hour) for ground (G) canopy (C) nets and harp traps (m.d. = missing data). Altitude (m)
630
720
900
1030
1220
1320 1850 2010
overall
Ground net hours
50.8
74.6
43.3
67.4
56.9
57.9
49.3
115.5
515.7
Canopy net hours
27.8
8.3
64.1
50.3
51.0
53.6
0
19.5
274.6
Harp trap hours
90.5
72.5
69.0
53.3
87.0
150.0 44.8
203.8
770.9
Ground net captures
7
53
1
6
5
0
0
0
72
Canopy net captures
5
9
13
25
5
5
NA
0
62
Ground capture success
0.14
0.71
0.02
0.09
0.09
0.00
0.00
0.00
1.00
Canopy capture success
0.18
1.08
0.20
0.50
0.10
0.09
NA
0.00
0.23
Net success ratio (C:G)
1.31
1.53
8.78
5.58
1.12
m.d.
NA
m.d.
0.23
a) Fruit bats
0
b) Animalivorous bats Ground net captures.
16
65
26
66
37
7
10
4
231
Canopy net captures.
2
0
2
2
1
1
NA
1
9
Harp trap captures
26
13
43
169
62
18
0
18
349
Ground capture success
0.32
0.87
0.60
0.98
0.65
0.12
0.20
0.04
0.45
Canopy capture success
0.07
0.00
0.03
0.04
0.02
0.02
NA
0.05
0.03
Harp trap capture success 0.29
0.18
0.62
3.17
0.71
0.12
0.00
0.09
0.45
Net success ratio (C:G)
m.d.
0.05
0.04
0.03
0.15
NA
1.48
0.07
0.23
S7. Non-Metric Multidimensional Scaling (NMDS) of assemblage composition of sampling sites based on CNESS dissimilarity index.
