Donelle Schwalm1*, Lisette P. Waits2 and Warren B. Ballard1

Little fox on the prairie: genetic structure and diversity throughout the distribution of a grassland carnivore in the United States

1 Department of Natural Resources Management, PO Box 42125, Texas Tech University,

Lubbock, Texas, USA

2 Department of Fish and Wildlife Resources, PO Box 441136, University of Idaho, Moscow,

Idaho, USA

Corresponding author: Donelle Schwalm, Department of Fisheries and Wildlife, Oregon State

University, Nash 104, Corvallis, OR 97331, E-mail: [email protected]

SUPPLEMENTAL INFORMATION Microsatellite PCR multiplex conditions

Microsatellite primers were divided into three multiplexes using the QIAGEN Multiplex Kit (QIAGEN). For tissue samples, PCR conditions (7μl final volume) for multiplex one were 1μM sample, 1X Master Mix, 0.5X Q solution, 0.05μM each primer for locus CXX173, 0.06μM each primer for loci CXX377 and FH2054, 0.09μM each primer for locus CXX20, 0.10μM each primer for loci CPH3 and CXX250, and 0.19μM each primer for locus CXX403. The PCR thermoprofile included initial denaturation for 15 min at 95ºC, 14 cycles of 30 s at 94ºC, 90 s at 55ºC (decreasing 0.3ºC per cycle to 50.8ºC) and 1 min at 72ºC, followed by 20 cycles of 30 s at 94ºC, 90 s at 51ºC and 1 min at 72ºC, then final elongation at 60ºC for 30 min. PCR conditions for multiplex two were 1μM sample, 1X Master Mix, 0.5X Q solution, 0.03μM each primer for locus CXX263, 0.08μM each primer for locus VVE2-111, 0.15μM each primer for locus CXX2062, 0.25μM each primer for locus VVE5-33, and 0.30μM each primer for locus CXX109. PCR conditions for multiplex three were 1μM sample, 1X Master Mix, 0.5X Q solution, 0.06μM each primer for locus VVE-M19, 0.08μM each primer for locus VVE3-131, 0.10μM each primer for locus VVE-M25, and 0.25μM each primer for locus VVE2-110. The PCR thermoprofile for multiplex two and three included initial denaturation for 15 min at 95ºC, 12 cycles of 30 s at 94ºC, 90 s at 53ºC (decreasing 0.3ºC per cycle to 49.4ºC) and 1 min at 72ºC, followed by 25 cycles of 30 s at 94ºC, 90 s at 47ºC and 1 min at 72ºC then final elongation at 60ºC for 30 min. We subsequently removed VVE-M25 from our analysis due to inconsistent peaks which rendered allele identification subjective. For blood and hair samples, we added 5 additional cycles at the 4th step of the thermoprofiles. mtDNA laboratory protocol Step 1, DNA amplification: PCR conditions (10μl final volume) were 1.5X MgCl2, 1.5X Goldtaq buffer, 0.3μM each primer (Thr-L15997 and Control H16401), 0.1 μM AmpliTaq Gold, 10μM DNTPs with a thermoprofile of initial denaturation for 10 min and 30 s at 95ºC, 20 cycles of 30 s starting at 51ºC and decreasing 0.2ºC per cycle, 1 min at 72ºC, 20 s at 95ºC, followed by 20 cycles of 30 s at 48ºC, then final elongation at 72ºC for 8 min. Step 2, DNA cleanup: following manufacturer protocol, we mixed 5 μl of product from step 1 with 2 μl of exosap, then ran the mixture at 37ºC for 15 min, 80ºC for 15 min, and 4ºC for 10 min. Step 3, sequencing: PCR conditions (10μl final volume) were 2X Big Dye, 2.4x sequencing buffer and 2X primer (Control H16401), with a thermoprofile of initial denaturation for 3 min at 9cºC, 25 cycles of 30 s at 95 ºC, 15 s at 50ºC and 4 min at 60ºC, followed by cool down at 4ºC for 10 min.

Rarefaction analysis for mtDNA haplotype detection

Figure S.1. Rarefaction curve showing saturation in the detection of new swift fox mitochondrial haplotypes at ~55 samples, with 95% confidence intervals.

Model parameterization for STRUCTURE and TESS Using the admixture model and correlated allele frequencies in the program STRUCTURE, we identified the optimal K value by first testing K = 1 - 15 with a burn-in period of 50,000 iterations and a total run time of 500,000 iterations for each of 10 replicates per candidate K value. If support for a specific value was not clear based on log likelihood values (i.e., if the graphed values did not show a clear peak at a given K value), we selected a K value using the delta K statistic described by Evanno et al. (2005).

To identify the optimal K value using TESS, we used a burn in period of 10,000 iterations and 50,000 total iterations for 20 replicates each K = 5 - 15. The estimated number of clusters stabilized at K = 6. Following program guidelines, we conducted the final analysis using 200 replicates at K = 7 with the same burn-in and total iterations previously defined and selected the top 20% of replicates (n = 40) for further use.

The program TESS constructs a neighborhood network which creates connections between samples. Average straight line post release movement distances (Fritts et al. 1997) and straight line natural dispersal distance (Shields 1987) for swift fox vary by region and sex, but typically average < 30 km (Allardyce & Sovada 2003; Kamler et al. 2004a). However, at least three long- distance movements have been recorded for swift fox: 411 km (Shaun Grassel, pers. com.), 298 km (Kevin Honness, pers. com.) and 191 km (Ausband & Moehrenschlager 2009). Thus, we removed neighborhood network connections greater than the mean for known long distance movements (300 km).

Results of group identification (K) using the program STRUCTURE

We detected 3 hierarchies of genetic structure using the program STRUCTURE. These results are presented by hierarchy, below.

Hierarchy one: In the first round of analysis, log-likelihood values did not clearly identify an optimal K value (Figure S.1); however, we detected support for two genetic groups using the delta K statistic (Figure S.2). Thus, in the first hierarchy of genetic structure, we concluded that there are two genetic groups. -23500.00

-24000.00

-24500.00

-25000.00

-25500.00

-26000.00

-26500.00

-27000.00

-27500.00

Figure S.2. Log-likelihood values for the first hierarchy of STRUCTURE analysis; no clear support for a specific K value was observed.

900.00 800.00 700.00 600.00 K

500.00 a t l

e 400.00 d 300.00 200.00 100.00 0.00

Figure S.3. Delta K values for genetic structure analysis, showing strong support for K = 2 and lower support for K = 5.

Hierarchy two: We next analyzed genetic structure in each of the two groups present in the first genetic structure hierarchy. We found each of these groups contained three unique genetic groups. Log- likelihood and delta K graphs for each of these analyses are shown below. -11700.00 K=1 K=2 K=3 K=4 K=5 K=6 -11800.00

-11900.00

-12000.00

-12100.00

-12200.00

-12300.00

-12400.00

-12500.00

Figure S.4. Log-likelihood values for genetic structure analysis of group 1, hierarchy 1; this result implies greatest support for K = 4.

140.00

120.00

100.00

K 80.00

a t l e 60.00 d

0.00 K=2 K=3 K=4 K=5 K=6

Figure S.5. Delta K values for genetic structure analysis of group 1, hierarchy 1; this result implies greatest support for K = 3. We chose this, the more conservative estimate of genetic structure, for further analysis. -12500.00 K=1 K=2 K=3 K=4 K=5 K=6

-13000.00

-13500.00

-14000.00

-14500.00

-15000.00

500.00 450.00 400.00 350.00 300.00 K

t 250.00 l e

d 200.00 150.00 100.00 50.00 0.00 K=2 K=3 K=4 K=5 K=6

Figure S.7. Delta K values for genetic structure analysis of group 2, hierarchy 1; this result implies greatest support for K = 3.

Hierarchy 3: We next assessed genetic structure in each of the six genetic groups identified in the second hierarchy. Of these, analysis indicated some groups were further subdivided and some represented a single group. Upon further investigation, some of the subdivisions detected lacked biological or ecological justification; in these instances, we concluded that a single genetic group was detected. In total we detected 10 genetic groups in the third hierarchy. The associated log- likelihood and delta K figures are presented below. If the highest log-likelihood was detected at K=1, delta K values were not estimated.

-5080.00 K=1 K=2 K=3 K=4 K=5 -5090.00

-5100.00

-5110.00

-5120.00

-5130.00

-5140.00

-5150.00

-5160.00

-5170.00

-5180.00

-4900.00 K=1 K=2 K=3 K=4 K=5 -5000.00

-5100.00

-5200.00

-5300.00

-5400.00

-5500.00

-5600.00

-5700.00

-5800.00

-5900.00

Figure S.9. Log-likelihood values for genetic structure analysis of group 2, hierarchy 2; this result implies greatest support for K = 1. -514.50 K=1 K=2 K=3 K=4 K=5

-515.00

-515.50

-516.00

-516.50

-517.00

-517.50

-5900.00 K=1 K=2 K=3 K=4 K=5 -5950.00

-6000.00

-6050.00

-6100.00

-6150.00

-6200.00

-6250.00

-6300.00

-6350.00

Figure S.11. Log-likelihood values for genetic structure analysis for genetic structure analysis of group 4, hierarchy 2; this result implies greatest support for K = 4. However, samples were highly admixed at q = 0.70 and lacked geographic cohesion. 40.00

25.00 K

t 20.00 l e d 15.00

0.00 K=2 K=3 K=4 K=5

Figure S.12. Delta K values for genetic structure analysis of group 4, hierarchy 2; this result implies greatest support for K = 3. Samples were highly admixed at q = 0.70 and lacked geographic cohesion; based on this and the results in Figure S.13, we concluded there was not support for further subdivision of this group.

-5900.00 K=1 K=2 K=3 K=4 K=5 -6000.00

-6100.00

-6200.00

-6300.00

-6400.00

-6500.00

-6600.00

Figure S.13. Log-likelihood values for genetic structure analysis for genetic structure analysis of group 5, hierarchy 2; this result implies greatest support for K = 3. 90.00 80.00 70.00 60.00 K

50.00 a t l

e 40.00 d 30.00 20.00 10.00 0.00 K=2 K=3 K=4 K=5

Figure S.14. Delta K values for genetic structure analysis of group 5, hierarchy 2; this result implies greatest support for K = 3. This group was subsequently split into 3 groups.

-1420.00 K=1 K=2 K=3 K=4 K=5 -1440.00

-1460.00

-1480.00

-1500.00

-1520.00

-1540.00

-1560.00

-1580.00

Figure S.15. Log-likelihood values for genetic structure analysis for genetic structure analysis of group 6, hierarchy 2; this result implies greatest support for K = 3. 35.00

K 20.00

a t l e 15.00 d

0.00 K=2 K=3 K=4 K=5

Figure S.16. Delta K values for genetic structure analysis of group 6, hierarchy 2; this result implies greatest support for K = 3. This group was subsequently split into 3 groups.

Table S1: Number of swift fox samples by the 8 hypothesized groups, sample type, sex and age class. Museum samples were muscle tissue collected from frozen carcasses and stored in lysis buffer. Sample Type Sex Class Age Class Museum Group Blood Hair Tissue Tissue Male Female Unknown Adult Juvenile Unknown 1 0 41 1 0 17 21 4 17 20 5 2 0 0 43 0 14 12 17 9 12 22 3 31 0 53 1 40 37 8 29 42 14 4 25 10 83 0 47 60 11 35 66 17 5 0 0 98 0 33 50 15 39 35 24 6 5 1 50 0 29 20 7 6 26 24 7 0 0 88 44 67 56 9 41 73 18 8 0 0 11 4 6 5 4 2 5 8 Total 61 52 427 49 253 261 75 178 279 132

Table S2. Size range and number (#) of alleles for each microsatellite locus used for analyzing swift fox samples. Size range # of Multiplex Locus (bp) Alleles FH20541 167-191 7 CPH32 150-160 6 CXX203 120-144 9 1 CXX1733 123-129 4 CXX2503 131-153 10 CXX3773 173-193 8 CXX4033 269-281 5 CXX1093 165-170 6 CXX2633 94-122 5 2 CXX20623 135-156 6 VVE2-1114 125-141 5 VVE5-334 190-226 9 VVE-M194 227-287 27 3 VVE2-1104 231-358 40 VVE3-1314 157-189 6 (Francisco et al. 1996) 1 (Fredholm et al. 1995) 2 (Ostrander et al. 1993) 3 (Cullingham et al. 2007) 4 Figure S.17. At the coarsest level of genetic structure (K = 2) we observed northern and southern groups with extensive admixture at their junction. These groups may represent a response to climatic influences (e.g., temperature, precipitation) along a latitudinal gradient. Differentiation along an environmental cline has been recorded in other North American canids (Geffen et al. 2004). Conversely, these groups may correspond to two expanding source populations, representing recovery from different refugia. Further research is necessary to explore these hypotheses. Figure S.18. Spatial arrangement of genetic groups identified by STRUCTURE at K = 6. These genetic groupings are similar to those observed for TESS analysis (K = 7), and largely correspond to hypothesized natural barriers.

Figure S.19. Geographic location of unique swift fox genetic groups identified using Bayesian clustering analysis. Shown are the results for STRUCTURE analysis at K = 10 and TESS analysis at K = 7, which represent emergent and long-term genetic structure, respectively. References Cullingham CI, Smeeton C, White BN (2007) Isolation and characterization of swift fox tetranucleotide microsatellite loci. Molecular Ecology Notes 7:160-162 Francisco LV, Langston AA, Mellersh CS, Neal CL, Ostrander EA (1996) A class of highly polymorphic tetranucleotide repeats for canine genetic mapping. Mammalian Genome 7: 359-362 Fredholm M, Wintero AK (1995) Variation of short tandem repeats within and between species belonging to the Canidae family. Mammalian Genome 6:11-18 Ostrander EA, Sprague GFJ, Rine J (1993) Identification and characterization of dinucleotide repeat (CA)n markers for genetic mapping in dog. Genomics 16:207-213