Kim and Holt Electronic Supplemental Material Oecologia 1

Kim and Holt Electronic Supplemental Material Oecologia 1

The direct and indirect effects of fire on the assembly of insect herbivore communities:

Examples from the Florida scrub habitat

Tania N. Kim & Robert D. Holt

Department of Biology, PO Box 118525

University of Florida

Gainesville, FL 32611-8525

USA

Correspondence to:

Tania N. Kim

Department of Biological Sciences

319 Stadium Drive

Florida State University

Tallahassee, FL 32306-4295

E-Mail:

Phone: 850-645-8575

Fax: 850-645-8447

Author Contributions: TNK and RDH formulated the idea, TNK developed methodology, conducted fieldwork, and performed statistical analyses, and TNK and RDH wrote the manuscript.

Kim and Holt Electronic Supplemental Material Oecologia 2

Fig. S1. a. The distribution of remaining scrub habitat in Florida. Today, less than 40% of pre-settlement scrub habitat remains in Florida, mostly in isolated fragments (Myers 1990). b. The main vegetation associations at Archbold Biological Station (ABS) in Highlands Co., FL, USA (Abrahamson 1984). Black-filled circles represent approximate locations of research plots along a chronosequence of time-since-fire in the scrubby flatwood habitat.

Kim and Holt Electronic Supplemental Material Oecologia 3

Fig. S2 Boxplots illustrating species-level differences in measured leaf traits and overall abundance of Quercus inopina, Q. geminata, Q. chapmanii. Tests for species-level treatment effects were calculated using one-way ANOVAs. Traits were averaged within and among plants in the same research plot to yield one average trait value per sampling session. Upper-case letters represent statistically significant (P<0.05) differences among oak species.

Kim and Holt Electronic Supplemental Material Oecologia 4

Fig. S3 Boxplots illustrating species-level and time-since fire (TSF) differences in measured plant traits of Quercus inopina, Q. geminata, Q. chapmanii. Tests for treatment effects (TSF and oak species) were calculated using two-way ANOVAs. TSF was statistically significant for all measured plant traits (all P0.02). Significant TSF and oak species interactions were observed in (a) and (b). Upper-case letters represent statistically significant (P<0.05) TSF differences among oak species.

Fig. S4 Biplot of NMDS analyses for herbivore community composition on Quercus inopina, Q. geminata, and Q. chapmanii. NMDS scores generated for herbivores found on all three oak species combined.

Kim and Holt Electronic Supplemental Material Oecologia 5

Table S1. Standardized path coefficients (or standardized partial regression coefficients) between endogenous and exogenous variables from each final structural equation model (Figure 1).

Leaf quality (LQ) / Plant architecture (PA) / Habitat structure (HS) / Herbivore abundance (AB) / Herbivore richness (RC) / Herbivore composition (CC)
A. Quercus inopina
TSF (linear) / 0 / 2.867 / 2.249 / 0.337 / 0.335 / 0.3
TSF2 (quadratic) / 0 / -2.396 / -2.257 / 0 / 0 / 0
Sampling session (SS) / 0 / 0 / 0 / -0.654 / -0.474 / -0.517
Landscape heterogeneity (LH) / 0 / 0 / 0 / 0 / 0 / 0
Leaf quality (LQ) / 0 / 0 / 0 / 0 / 0 / 0
Plant architecture (PA) / 0 / 0 / 0 / 0 / 0 / 0
Habitat structure (HS) / 0 / 0 / 0 / 0.211 / 0.267 / 0
B. Quercus geminata
TSF (linear) / -0.276 / 2.77 / 2.25 / 0 / 0 / 0
TSF2 (quadratic) / 0 / -2.385 / -2.266 / 0 / 0 / 0
Sampling session (SS) / -0.424 / 0 / 0 / -0.369 / -0.393 / -0.341
Landscape heterogeneity (LH) / 0 / 0 / 0 / -0.280 / -0.205 / 0.258
Leaf quality (LQ) / 0 / 0 / 0 / 0 / 0 / 0
Plant architecture (PA) / 0 / 0 / 0 / 0 / 0 / 0
Habitat structure (HS) / 0 / 0 / 0 / 0 / 0 / 0
C. Quercus chapmanii
TSF (linear) / 0 / 2.368 / 2.327 / 0 / 0 / 0
TSF2 (quadratic) / -0.347 / -1.726 / -2.311 / 0 / 0 / 0
Sampling session (SS) / -0.372 / 0 / 0 / -0.188 / -0.317 / 0
Landscape heterogeneity (LH) / 0 / 0 / 0 / 0 / 0 / 0
Leaf quality (LQ) / 0 / 0 / 0 / 0 / 0 / 0
Plant architecture (PA) / 0 / 0 / 0 / 0 / 0 / 0
Habitat structure (HS) / 0 / 0 / 0 / 0 / 0 / 0

Kim and Holt Electronic Supplementary Material Oecologia 6

Table S2. Standardized model results for the direct, indirect, and total effect of predictor variables on (a) herbivore abundance, (b) richness, and (c) community composition for herbivore assemblages on Q. inopina, Q. geminata, and Q. chapmanii. Coefficients in the table represent the standardized path coefficients (or standardized partial regression coefficients) indicating the strength of each relationships in final SE model. Indirect effects were calculated by multiplying the standardized path coefficients in each of the pathway and total effects were the sum of the direct and indirect effects (Figure 1, Table S1).

Quercus inopina / Quercus geminata / Quercus chapmanii
Direct / Indirect / Total / Direct / Indirect / Total / Direct / Indirect / Total
A. Herbivore abundance
·  Time-since-fire(linear) / 0.337 / 0.475 / 0.812 / 0 / 0 / 0 / 0 / 0 / 0
·  Time-since-fire2(quadratic) / 0 / -0.477 / -0.477 / 0 / 0 / 0 / 0 / 0 / 0
·  Sampling session / -0.654 / 0 / -0.654 / -0.369 / 0 / -0.369 / -0.188 / 0 / -0.188
·  Landscape heterogeneity / 0 / 0 / 0 / -0.280 / 0 / -0.280 / 0 / 0 / 0
·  Habitat structure / 0.211 / 0 / 0.211 / 0 / 0 / 0 / 0 / 0 / 0
·  Plant architecture / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0
·  Leaf quality / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0
B. Herbivore richness
·  Time-since-fire(linear) / 0.335 / 0.601 / 0.936 / 0 / 0 / 0 / 0 / 0 / 0
·  Time-since-fire2(quadratic) / 0 / -0.603 / -0.603 / 0 / 0 / 0 / 0 / 0 / 0
·  Sampling session / -0.474 / 0 / -0.474 / -0.393 / 0 / -0.393 / -0.317 / 0 / -0.317
·  Landscape heterogeneity / 0 / 0 / 0 / -0.205 / 0 / -0.205 / 0 / 0 / 0
·  Habitat structure / 0.267 / 0 / 0.267 / 0 / 0 / 0 / 0 / 0 / 0
·  Plant architecture / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0
·  Leaf quality / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0
C. Herbivore community composition
·  Time-since-fire(linear) / 0.300 / 0 / 0.300 / 0 / 0 / 0 / 0 / 0 / 0
·  Time-since-fire2(quadratic) / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0
·  Sampling session / -0.517 / 0 / -0.517 / -0.341 / 0 / -0.341 / 0 / 0 / 0
·  Landscape heterogeneity / 0 / 0 / 0 / 0.258 / 0 / 0.258 / 0 / 0 / 0
·  Habitat structure / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0
·  Plant architecture / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0
·  Leaf quality / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 0

Kim and Holt Electronic Supplementary Material Oecologia 7

Table S3. Significant correlation coefficient values (double-headed arrows) from each final structural equation model (Figure 1).

A. Quercus inopina / TSF / AB / RC / CC
·  Linear time-since-fire (TSF) / - / - / - / -
·  Quadratic time-since-fire (TSF2) / 0.97 / - / - / -
·  Landscape heterogeneity (LH) / 0.17 / - / - / -
·  Herbivore abundance (AB) / - / - / - / -
·  Herbivore richness (RC) / - / 0.8 / - / -
·  Herbivore community composition (CC) / - / 0.6 / 0.39 / -
B. Quercus geminata / TSF / AB / RC / CC
·  Linear time-since-fire (TSF) / - / - / - / -
·  Quadratic time-since-fire (TSF2) / 0.97 / - / - / -
·  Landscape heterogeneity (LH) / 0.17 / - / - / -
·  Herbivore abundance (AB) / - / - / - / -
·  Herbivore richness (RC) / - / 0.83 / - / -
·  Herbivore community composition (CC) / - / -0.3 / - / -
C. Quercus chapmanii / TSF / AB / RC / CC
·  Linear time-since-fire (TSF) / 0.97 / - / - / -
·  Quadratic time-since-fire (TSF2) / 0.17 / - / - / -
·  Landscape heterogeneity (LH) / - / - / - / -
·  Herbivore abundance (AB) / - / - / - / -
·  Herbivore richness (RC) / - / 0.81 / - / -
·  Herbivore community composition (CC) / - / 0.87 / 0.51 / -