Ecological Archives E096-253-A1
Nicholas A. Pardikes, Arthur M. Shapiro, Lee A. Dyer, and Matthew L. Forister. 2015. Global weather and local butterflies: variable responses to a large-scale climate pattern along an elevational gradient. Ecology 96:2891–2901. http://dx.doi.org/10.1890/15-0661.1
Appendix A. Supplementary information, including detailed monitoring site information, species used in this analysis, presence data over the 23-year monitoring period, correlation matrix, additional ANOVA tables, and results from additional path analyses.
Table A1. A table revealing the sources for local weather values and the months in which data was missing.
Site |
Elevation (m) |
Weather Station |
Missing Data filled in with PPCA |
Suisun Marsh |
0–1 |
Fairfield, 042934 (38.2667, -122.06667) |
N/A |
North Sacramento |
8 |
Sac. FAA Airport, 047630 (38.5069, -121.5) |
N/A |
West Sacramento |
9 |
Sac. 5 ESE, 047633 (38.55556, -121.95) |
N/A |
Rancho Cordova |
18 |
PRISM (39.6241, -121.2777) |
N/A |
Gates Canyon |
190–600 |
Vacaville, 049200 (38.416667, -121.95) |
1988 (April, May, June), 1989 (Nov.), 1990 (March, Dec.), 1994 (Oct.), 1998 (Aug.) |
Washington |
850–1,200 |
Nevada City, 6316 (39.26, -121.02) |
N/A |
Sierra Valley |
1,500 |
Sierraville Ranger Station, 048218 (39.58333, -120.36667) |
N/A |
Lang Crossing |
1,500–1,700 |
PRISM (39.315, -120.662) |
N/A |
Donner Pass |
2,000–2,200 |
Sierra Snow Lab, 049998 (39.326, -120.367) |
N/A |
Castle Peak |
2,400–2,775 |
PRISM (39.3395, -120.3474) |
N/A |
Notes: Sites are ordered from low to high elevation. Latitude and longitude are provided in parentheses for each weather station. Missing values were filled in using a Probabilistic Principle Coordinates Analysis (PPCA) in the “pcaMethods” package in R (Stacklies et al. 2012). N/A values represent sites that did not have any missing values.
Table A2. A list of the 28 butterfly species and their family used in this analysis.
Species |
Family |
Adelpha bredowii californica |
Nymphalidae |
Atalopedes campestris |
Hesperiidae |
Celastrina ladon echo |
Lycaenidae |
Colias eurytheme |
Pieridae |
Danaus plexippus |
Nymphalidae |
Erynnis persius |
Hesperiidae |
Euchloe ausonides |
Pieridae |
Hylephila phyleus |
Hesperiidae |
Junonia coenia |
Nymphalidae |
Limenitis lorquini |
Nymphalidae |
Lycaena helloides |
Lycaenidae |
Nymphalis antiopa |
Nymphalidae |
Nymphalis californica |
Nymphalidae |
Nymphalis milberti |
Nymphalidae |
Ochlodes sylvanoides |
Hesperiidae |
Papilio rutulus |
Papilionidae |
Papilio zelicaon |
Papilionidae |
Phyciodes mylitta |
Nymphalidae |
Pieris rapae |
Pieridae |
Plebejus acmon |
Lycaenidae |
Pontia protodice |
Pieridae |
Pyrgus communis |
Hesperiidae |
Satyrium sylvinus |
Lycaenidae |
Strymon melinus |
Lycaenidae |
Vanessa annabella |
Nymphalidae |
Vanessa atalanta |
Nymphalidae |
Vanessa cardui |
Nymphalidae |
Vanessa virginiensis |
Nymphalidae |
Table A3. Number of years that each species was seen over the 23-year monitoring period.
Species |
SM |
NS |
WS |
RC |
GC |
WA |
SV |
LC |
DP |
CP |
A. bredowii californica |
14 |
1 |
2 |
14 |
20 |
17 |
9 |
18 |
18 |
8 |
Atalopedes campestris |
19 |
21 |
20 |
20 |
19 |
14 |
8 |
3 |
1 |
1 |
Celastrina ladon echo |
3 |
4 |
1 |
3 |
16 |
15 |
7 |
17 |
18 |
11 |
Colias eurytheme |
21 |
18 |
20 |
19 |
19 |
19 |
16 |
15 |
14 |
6 |
Danaus plexippus |
19 |
21 |
21 |
19 |
19 |
18 |
19 |
20 |
16 |
11 |
Erynnis persius |
2 |
4 |
2 |
4 |
15 |
17 |
3 |
17 |
8 |
2 |
Euchloe ausonides |
19 |
16 |
15 |
11 |
17 |
7 |
17 |
4 |
1 |
7 |
Hylephila phyleus |
19 |
20 |
17 |
18 |
21 |
14 |
3 |
3 |
4 |
7 |
Junonia coenia |
21 |
19 |
17 |
19 |
22 |
20 |
12 |
19 |
15 |
9 |
Limenitis lorquini |
2 |
13 |
22 |
15 |
17 |
17 |
13 |
19 |
15 |
9 |
Lycaena helloides |
20 |
19 |
21 |
10 |
15 |
6 |
16 |
14 |
8 |
3 |
Nymphalis antiopa |
16 |
18 |
19 |
21 |
17 |
21 |
14 |
19 |
21 |
12 |
Nymphalis californica |
13 |
11 |
9 |
18 |
18 |
19 |
17 |
19 |
20 |
13 |
Nymphalis milberti |
2 |
8 |
1 |
3 |
1 |
2 |
5 |
2 |
15 |
13 |
Ochlodes sylvanoides |
16 |
16 |
8 |
19 |
15 |
18 |
16 |
14 |
16 |
10 |
Papilio rutulus |
18 |
17 |
18 |
17 |
19 |
19 |
16 |
16 |
19 |
10 |
Papilio zelicaon |
19 |
18 |
19 |
19 |
18 |
18 |
11 |
10 |
18 |
11 |
Phyciodes mylitta |
18 |
18 |
21 |
21 |
19 |
18 |
18 |
20 |
18 |
7 |
Pieris rapae |
15 |
18 |
20 |
21 |
17 |
17 |
15 |
16 |
18 |
14 |
Plebejus acmon |
19 |
21 |
21 |
19 |
19 |
16 |
16 |
18 |
16 |
11 |
Pontia protodice |
13 |
14 |
17 |
18 |
11 |
9 |
20 |
11 |
18 |
8 |
Pyrgus communis |
18 |
19 |
16 |
21 |
16 |
20 |
19 |
19 |
14 |
13 |
Satyrium sylvinus |
1 |
17 |
15 |
6 |
18 |
14 |
9 |
14 |
15 |
3 |
Strymon melinus |
19 |
17 |
19 |
17 |
19 |
17 |
14 |
10 |
13 |
6 |
Vanessa annabella |
18 |
21 |
20 |
20 |
21 |
17 |
17 |
16 |
19 |
15 |
Vanessa atalanta |
19 |
19 |
21 |
20 |
20 |
12 |
6 |
9 |
8 |
3 |
Vanessa cardui |
21 |
20 |
21 |
22 |
20 |
16 |
20 |
15 |
18 |
16 |
Vanessa virginiensis |
16 |
19 |
20 |
19 |
16 |
18 |
16 |
16 |
18 |
12 |
Notes: A.M.S. visited each site multiple times throughout the year; therefore years that the butterfly was absent from a particular site are meaningful absences.
Table A4. Pearson’s Correlation Coefficients for the variables used in theses analyses.
|
Visits |
N |
N(t-1) |
MinT |
MaxT |
Precip |
SSTA |
Visits |
1 |
0.0489 |
0.066 |
-0.11 |
-0.009 |
0.0143 |
0.001 |
N |
- |
1 |
0.812 |
0.020 |
0.026 |
0.0038 |
0.034 |
N(t-1) |
- |
- |
1 |
0.0007 |
-0.015 |
0.045 |
-0.024 |
MinT |
- |
- |
- |
1 |
0.439 |
0.245 |
0.155 |
MaxT |
- |
- |
- |
- |
1 |
-0.408 |
-0.216 |
Precip |
- |
- |
- |
- |
- |
1 |
0.129 |
SSTA |
- |
- |
- |
- |
- |
- |
1 |
Notes: SSTA and local weather variables are z-standardized.
Table A5. Results from χ² Type III analyses of deviance for the GLMM, Table 1, Model 1.
Fixed-Effect |
χ² |
df |
Pr(>χ²) |
|
(Intercept) |
668.5707 |
1 |
<0.0001 |
*** |
N (t-1) |
8557.3922 |
1 |
<0.0001 |
*** |
SSTA |
237.5695 |
1 |
<0.0001 |
*** |
MaxT |
84.1841 |
1 |
<0.0001 |
*** |
Precip |
6.6673 |
1 |
0.010 |
** |
MinT |
3.9659 |
1 |
0.05 |
* |
Notes: The analysis of deviance was performed in the R package “car” (Fox et al. 2015). Variables are ordered from highest to lowest χ² values. The main effect of interest (SSTA) is shown in bold. *indicates significance at P < 0.05.
Table A6. Results from χ² Type III analyses of deviance for the GLMM, Table 1, Model 1 (Resident Data).
Fixed-Effect |
χ² |
df |
Pr(>χ²) |
|
(Intercept) |
364.6199 |
1 |
<0.0001 |
*** |
N (t-1) |
5181.5489 |
1 |
<0.0001 |
*** |
SSTA |
91.2648 |
1 |
<0.0001 |
*** |
MaxT |
71.7935 |
1 |
<0.0001 |
*** |
Precip |
8.1446 |
1 |
0.004 |
** |
MinT |
0.5231 |
1 |
0.47 |
|
Notes: The analysis of deviance was performed in the R package “car” (Fox et al. 2015). This model corresponds to Model 1, Table 1, but only uses resident data. Variables are ordered from highest to lowest χ² values. The main effect of interest (SSTA) is shown in bold. *indicates significance at P < 0.05.
Table A7. Results from χ² Type III analyses of deviance for the GLMM, Table 1, Model 1 (Non-Resident Data)
Fixed-Effect |
χ² |
df |
Pr(>χ²) |
|
(Intercept) |
49.6655 |
1 |
<0.0001 |
*** |
SSTA |
191.7437 |
1 |
<0.0001 |
*** |
N (t-1) |
126.6247 |
1 |
<0.0001 |
*** |
MaxT |
7.891 |
1 |
0.005 |
** |
Precip |
5.3939 |
1 |
0.020 |
* |
MinT |
0.9715 |
1 |
0.324 |
|
Notes: The analysis of deviance was performed in the R package “car” (Fox et al. 2015). This model corresponds to Model 1, Table 1, but only uses non-resident data. Variables are ordered from highest to lowest χ² values. The main effect of interest (SSTA) is shown in bold. *indicates significance at P < 0.05.
Table A8. Results from χ² Type III analyses of deviance for the GLMM, Table 1, Model 1 (Valley Data)
Fixed-Effect |
χ² |
df |
Pr(>χ²) |
|
(Intercept) |
204.7118 |
1 |
<0.0001 |
*** |
N (t-1) |
2697.164 |
1 |
<0.0001 |
*** |
SSTA |
172.0705 |
1 |
<0.0001 |
*** |
MaxT |
61.315 |
1 |
<0.0001 |
*** |
Precip |
9.8796 |
1 |
0.002 |
** |
MinT |
0.0999 |
1 |
0.752 |
|
Notes: The analysis of deviance was performed in the R package “car” (Fox et al. 2015). This model corresponds to Model 1, Table 1, but only uses data from the five valley sites. Variables are ordered from highest to lowest χ² values. The main effect of interest (SSTA) is shown in bold. *indicates significance at P < 0.05.
Table A9. Results from χ² Type III analyses of deviance for the GLMM, Table 1, Model 1 (Mountain Data)
Fixed-Effect |
χ² |
df |
Pr(>χ²) |
|
(Intercept) |
283.8398 |
1 |
<0.0001 |
*** |
N (t-1) |
1956.8224 |
1 |
<0.0001 |
*** |
SSTA |
42.3341 |
1 |
<0.0001 |
*** |
MaxT |
10.7295 |
1 |
0.0011 |
** |
MinT |
9.5517 |
1 |
0.0020 |
** |
Precip |
1.4842 |
1 |
0.2231 |
|
Notes: The analysis of deviance was performed in the R package “car” (Fox et al. 2015). This model corresponds to Model 1 Table 1, but only uses data from the five mountain sites. Variables are ordered from highest to lowest χ² values. The main effect of interest (SSTA) is shown in bold. *indicates significance at P < 0.05.
Table A10. Results from χ² Type III analyses of deviance for the GLMM, Table 1, Model 2.
Fixed-Effect |
χ² |
df |
Pr(>χ²) |
|
(Intercept) |
582.2475 |
1 |
<0.0001 |
*** |
N (t-1) |
8539.4405 |
1 |
<0.0001 |
*** |
MaxT |
81.617 |
1 |
<0.0001 |
*** |
Site |
70.0246 |
9 |
<0.0001 |
*** |
SSTA × Site |
33.7933 |
9 |
<0.0001 |
*** |
SSTA |
10.9411 |
1 |
0.0009 |
*** |
Precip |
6.9186 |
1 |
0.009 |
** |
MinT |
4.0445 |
1 |
0.0443 |
* |
Notes: The analysis of deviance was performed in the R package “car” (Fox et al. 2015). Variables are ordered from highest to lowest χ² values. The interaction of interest (SSTA × Site) is shown in bold. *indicates significance at P < 0.05.
Table A11. Results from χ² Type III analyses of deviance for the GLMM, Table 1, Model 3.
Fixed-Effect |
χ² |
df |
Pr(>χ²) |
|
(Intercept) |
2954.7533 |
1 |
<0.0001 |
*** |
N (t-1) |
8723.557 |
1 |
<0.0001 |
*** |
Species |
1245.2295 |
27 |
<0.0001 |
*** |
SSTA × Species |
627.1538 |
27 |
<0.0001 |
*** |
MaxT |
82.9208 |
1 |
<0.0001 |
*** |
Precip |
6.9124 |
1 |
0.00856 |
** |
MinT |
4.2628 |
1 |
0.03896 |
* |
SSTA |
1.465 |
1 |
0.22613 |
|
Notes: The analysis of deviance was performed in the R package “car” (Fox et al. 2015). Variables are ordered from highest to lowest χ² values. The interaction of interest (SSTA × Species) is shown in bold. *indicates significance at P < 0.05.
Table A12. Results from Type III analyses of deviance for the GLMM Table 1, Model 4.
Fixed-Effect |
LR χ² |
df |
Pr(>χ²) |
|
Species × Site |
5237.3 |
243 |
<0.0001 |
*** |
Site |
909.9 |
9 |
<0.0001 |
*** |
Species |
720.1 |
27 |
<0.0001 |
*** |
N (t-1) |
436.6 |
1 |
<0.0001 |
*** |
SSTA × Species × Site |
268.4 |
243 |
0.1259 |
|
SSTA × Species |
74.3 |
27 |
2.71E-06 |
*** |
MaxT |
55.3 |
1 |
1.03E-13 |
*** |
SSTA × Site |
11.9 |
9 |
0.2186 |
|
SSTA |
2.6 |
1 |
0.1067 |
|
MinT |
0.5 |
1 |
0.4847 |
|
Precip |
0.2 |
1 |
0.6765 |
|
Notes: The analysis of deviance was performed in the R package “car” (Fox et al. 2015). Variables are ordered from highest to lowest χ² values. The interaction of interest (SSTA × Species × Site) is shown in bold. *indicates significance at P < 0.05.
Table A13. Results from the path analysis in Fig. 4.
Path |
Estimate |
SE |
t value |
P value |
Visits → #Pos. Sightings |
0.38 |
0.01 |
34.70 |
< .0001 * |
Year → Visits |
0.18 |
0.01 |
14.91 |
< .0001 * |
SSTA → MintT |
0.15 |
0.01 |
12.36 |
< .0001 * |
Precip → Visits |
0.12 |
0.02 |
8.17 |
< .0001 * |
MaxT → Visits |
0.11 |
0.02 |
6.74 |
< .0001 * |
SSTA → Precip |
0.08 |
0.01 |
6.75 |
< .0001 * |
SSTA → #Pos. Sightings |
0.04 |
0.01 |
3.52 |
0.0004 * |
Year → Precip |
0.04 |
0.01 |
2.93 |
0.0034 * |
MaxT → Pos. Sightings |
0.04 |
0.02 |
2.12 |
0.0342 * |
Year → MinT |
0.03 |
0.01 |
2.37 |
0.0180 * |
Precip → #Pos. Sightings |
0.01 |
0.01 |
0.70 |
0.4824 |
MinT → #Pos. Sightings |
0.01 |
0.02 |
0.60 |
0.5506 |
SSTA → #Pos. Sightings (indirect) |
-0.02 |
0.00 |
-4.27 |
< .0001 * |
Year → MaxT |
-0.07 |
0.01 |
-5.81 |
< .0001 * |
Year → #Pos. Sightings |
-0.09 |
0.01 |
-7.75 |
< .0001 * |
SSTA → MaxT |
-0.19 |
0.01 |
-15.90 |
< .0001 * |
MinT → Visits |
-0.20 |
0.02 |
-12.99 |
< .0001 * |
Notes: Displays model paths and their associated coefficients from Figure 4, including paths that were omitted from the figure for simplicity sake. Paths omitted from Figure 4 are shown in bold. Direct (SSTA → #Pos. Sightings) and indirect (SSTA → #Pos. Sightings (indirect)) effects of SSTA on the abundance of butterflies are shown in italics. Paths are ordered from most positive to most negative path coefficients. *indicates significance at P < 0.05.
Table A14. Results from the path analyses in Fig. 4 for each individual species.
Species |
Direct |
P value |
Total Indirect |
P value |
V. cardui |
0.41 |
< .0001 * |
0.02 |
0.5248 |
P. protodice |
0.27 |
< .0001 * |
-0.04 |
0.1611 |
V. virginiensis |
0.14 |
0.03 * |
-0.03 |
0.2215 |
J. coenia |
0.13 |
0.001 * |
-0.05 |
0.056 |
A. campestris |
0.12 |
0.003 * |
-0.09 |
0.0014 * |
V. atalanta |
0.11 |
0.0122 * |
-0.03 |
0.2995 |
C. eurytheme |
0.10 |
0.0033 * |
-0.07 |
0.0123 * |
P. acmon |
0.10 |
0.06 |
-0.04 |
0.1044 |
N. milberti |
0.08 |
0.17 |
-0.04 |
0.1588 |
E. ausonides |
0.07 |
0.25 |
-0.04 |
0.0719 |
P. rutulus |
0.06 |
0.27 |
-0.05 |
0.0607 |
D. plexippus |
0.06 |
0.32 |
-0.01 |
0.6748 |
V. annabella |
0.05 |
0.35 |
0.01 |
0.5815 |
S. melinus |
0.05 |
0.21 |
-0.05 |
0.0436 * |
P. rapae |
0.04 |
0.08 |
-0.05 |
0.0557 |
L. helloides |
0.04 |
0.53 |
-0.05 |
0.0604 |
P. communis |
0.04 |
0.40 |
-0.07 |
0.0093 * |
S. sylvinus |
0.03 |
0.65 |
-0.09 |
0.0024 * |
H. phyleus |
0.02 |
0.61 |
-0.06 |
0.0153 * |
P. myllitta |
0.01 |
0.90 |
-0.04 |
0.1417 |
L. lorquini |
-0.01 |
0.89 |
-0.01 |
0.6342 |
P. zelicaon |
-0.03 |
0.60 |
0.01 |
0.7054 |
E. persius |
-0.03 |
0.67 |
0.04 |
0.0952 |
N. antiopa |
-0.05 |
0.42 |
-0.04 |
0.1639 |
C. ladon echo |
-0.06 |
0.39 |
0.05 |
0.0559 |
N. californica |
-0.09 |
0.18 |
0.01 |
0.7184 |
A. bredowii |
-0.10 |
0.17 |
0.08 |
0.008 * |
O. sylvanoides |
-0.13 |
0.06 |
0.04 |
0.1006 |
Notes: Species are ordered from highest to lowest direct SSTA estimate. The total indirect SSTA estimates coincide with the dashed line from FIG. 4. χ² values of model fit were all the same for each species (Pr (>χ²=0.7195)). *indicates significance at P < 0.05.
Table A15. Results from the path analyses in Fig. 4 when performed for each individual site.
Site |
Pr (>χ²) |
Direct |
P value |
Total Indirect |
P value |
CP |
< .0001 |
0.096 |
0.017 * |
0.0004 |
0.9767 |
DP |
< .0001 |
0.032 |
0.4279 |
0.0203 |
0.0533 * |
LC |
0.0321 |
0.043 |
0.3065 |
-0.0238 |
0.1324 |
SV |
0.1494 |
0.035 |
0.5475 |
-0.0022 |
0.9604 |
WA |
0.3216 |
0.040 |
0.3745 |
-0.007 |
0.7637 |
GC |
0.0049 |
0.090 |
0.0417 * |
-0.005 |
0.8163 |
RC |
< .0001 |
0.046 |
0.2951 |
0.0137 |
0.4679 |
WS |
< .0001 |
0.021 |
0.6227 |
-0.002 |
0.8979 |
NS |
0.0001 |
0.043 |
0.321 |
-0.006 |
0.7307 |
SM |
0.1201 |
0.043 |
0.3597 |
-0.0126 |
0.6183 |
|
|||||
Notes: Sites are ordered from highest to lowest elevation. Higher p values represent greater support for the overall path model. *indicates significance at P < 0.05.
Literature Cited
Fox, J., S. Weisberg, D. Adler, D. Bates, G. Baud-Bovy, S. Ellison, D. Firth, M. Friendly, G. Gorjanc, S. Graves, R. Heiberger, R. Laboissiere, G. Monette, D. Murdoch, H. Nilsson, D. Ogle, B. Ripley, W. Venables, A. Zeileis, and R-Core. 2015. car: Companion to Applied Regression.
Stacklies, W., H. Redestig, M. Scholz, D. Walther, and J. Selbig. 2007. pcaMethods—a bioconductor package providing PCA methods for incomplete data. Bioinformatics 23:1164–1167.