diff --git a/internal/pipeline/candidate_fusion_test.go b/internal/pipeline/candidate_fusion_test.go
new file mode 100644
index 0000000..ae33cfd
--- /dev/null
+++ b/internal/pipeline/candidate_fusion_test.go
@@ -0,0 +1,64 @@
+package pipeline
+
+import "testing"
+
+func TestFuseCandidatesDedup(t *testing.T) {
+	primary := []Candidate{
+		{
+			ID:          1,
+			CenterHz:    100.0e6,
+			BandwidthHz: 20000,
+			Evidence: []LevelEvidence{
+				{Level: AnalysisLevel{Name: "surveillance"}},
+			},
+		},
+	}
+	derived := []Candidate{
+		{
+			ID:          -1,
+			CenterHz:    100.0e6 + 2000,
+			BandwidthHz: 25000,
+			Evidence: []LevelEvidence{
+				{Level: AnalysisLevel{Name: "surveillance-lowres"}},
+			},
+		},
+	}
+	fused := FuseCandidates(primary, derived)
+	if len(fused) != 1 {
+		t.Fatalf("expected 1 fused candidate, got %d", len(fused))
+	}
+	if got := CandidateEvidenceLevelCount(fused[0]); got != 2 {
+		t.Fatalf("expected 2 evidence levels after fuse, got %d", got)
+	}
+}
+
+func TestFuseCandidatesSingleVsMultiResolution(t *testing.T) {
+	primary := []Candidate{
+		{
+			ID:          1,
+			CenterHz:    101.0e6,
+			BandwidthHz: 12000,
+			Evidence: []LevelEvidence{
+				{Level: AnalysisLevel{Name: "surveillance"}},
+			},
+		},
+	}
+	single := FuseCandidates(primary, nil)
+	if len(single) != 1 {
+		t.Fatalf("expected single-resolution to keep 1 candidate, got %d", len(single))
+	}
+	derived := []Candidate{
+		{
+			ID:          -2,
+			CenterHz:    101.3e6,
+			BandwidthHz: 15000,
+			Evidence: []LevelEvidence{
+				{Level: AnalysisLevel{Name: "surveillance-lowres"}},
+			},
+		},
+	}
+	multi := FuseCandidates(primary, derived)
+	if len(multi) != 2 {
+		t.Fatalf("expected multi-resolution to keep 2 candidates, got %d", len(multi))
+	}
+}
diff --git a/internal/pipeline/scheduler_test.go b/internal/pipeline/scheduler_test.go
index ffa2080..e6fa4f4 100644
--- a/internal/pipeline/scheduler_test.go
+++ b/internal/pipeline/scheduler_test.go
@@ -144,6 +144,37 @@ func TestScheduleCandidatesPriorityBoost(t *testing.T) {
 	}
 }
 
+func TestScheduleCandidatesEvidenceBoost(t *testing.T) {
+	policy := Policy{MaxRefinementJobs: 2, MinCandidateSNRDb: 0}
+	single := Candidate{
+		ID:          1,
+		SNRDb:       8,
+		BandwidthHz: 12000,
+		Evidence: []LevelEvidence{
+			{Level: AnalysisLevel{Name: "surveillance"}},
+		},
+	}
+	multi := Candidate{
+		ID:          2,
+		SNRDb:       8,
+		BandwidthHz: 12000,
+		Evidence: []LevelEvidence{
+			{Level: AnalysisLevel{Name: "surveillance"}},
+			{Level: AnalysisLevel{Name: "surveillance-lowres"}},
+		},
+	}
+	plan := BuildRefinementPlan([]Candidate{single, multi}, policy)
+	if len(plan.Ranked) < 2 {
+		t.Fatalf("expected ranked candidates, got %d", len(plan.Ranked))
+	}
+	if plan.Ranked[0].Candidate.ID != multi.ID {
+		t.Fatalf("expected multi-level candidate to rank first, got %+v", plan.Ranked[0])
+	}
+	if plan.Ranked[0].Breakdown == nil || plan.Ranked[0].Breakdown.EvidenceScore <= 0 {
+		t.Fatalf("expected evidence score to be populated, got %+v", plan.Ranked[0].Breakdown)
+	}
+}
+
 func TestBuildRefinementPlanPriorityStats(t *testing.T) {
 	policy := Policy{MaxRefinementJobs: 1, MinCandidateSNRDb: 0}
 	cands := []Candidate{