Variants called using GATK from samples of Nextera and TruSeq exome enrichment kits

[bioinf newbie]

I am doing variant calling for exomes generated from Illumina HiSeq 2000, with Nextera Exome Enrichment kit.
Earlier I used the same variant calling pipeline to call variants from sequences based on TruSeq Exome Enrichment kit.
Now with the Nextera Exome Enrichment kit, I find that the number of variants called using GATK is reduced to almost half as compared to TruSeq.
I am not able to reach a conclusion as to why this drastic reduction in number.

TruSeq sample Nextera sample
Mean coverage 40X 47.5X
GATK variants 53917 27596

Here is the GATK command I used for calling variants for both TruSeq and Nextera samples:
java -Xmx2g -Djava.io.tmpdir=temp -jar $gatk_jar -T UnifiedGenotyper -I $in -R $ucsc_hg19_fasta -D:name,VCF $dbsnp146_hg19_vcf -o $o -dcov 1000 -A AlleleBalance -A BaseCounts -A VariantType -baq CALCULATE_AS_NECESSARY -stand_call_conf 30.0 -stand_emit_conf 10.0 -glm BOTH -L $nextera_target_bed(or $truseq_target_bed )

[HESmith]

A few possible explanations:

1) higher error rate in the sequence data from the TruSeq sample, producing false-positive variant calls. You can use FastQC to assess the base quality metrics of the two data sets.

2) capture of highly polymorphic loci in the TruSeq but not Nextera enrichment. Compare the covered intervals with Bedtools, and see if there's a large difference in variant frequency between the shared vs unique regions.