Design Experiments From Reported Results

 Long Analysis Plan Copied

Experimental Design: 3×6 Mesocosm Test of Genetic vs Demographic Rescue

Core question: Does adding genetically diverse immigrants (genetic rescue) outperform adding numerically equivalent low-diversity individuals (demographic augmentation), or any addition at all (control), on population growth, adaptive potential, and genomic dynamics over 6 generations?

The design is powered by the temporal-autocovariance framework of Buffalo & Coop, which shows that neutral allele-frequency covariance across generations [Cov(Δp_t, Δp_s)] rises under linked selection in proportion to the additive genetic variance for fitness, enabling joint estimation of V_A(1) and N_e from time-series genotypes via a method-of-moments estimator validated by forward Wright–Fisher simulation across N = 100–1000 Buffalo & Coop temporal autocovariance framework. This matters because real-world augmentation often fails: in Wisconsin martens, 68 translocated individuals contributed minimally to recruitment (0.57 ± 0.24) and juvenile survival (0.39 ± 0.21), with persistence driven by intrinsic vital rates Marten augmentation rescue study.

Design Specification

• Units: 18 mesocosms (3 treatments × 6 replicates), randomized in a blocked layout; experimental unit = mesocosm.
• Founder population: N = 100 census adults per mesocosm, drawn from a single inbred source stock to create low starting V_A.
• Treatments (G1 pulse, M = 10): (1) Genetic rescue — 10 unrelated immigrants from a divergent outbred stock; (2) Demographic augmentation — 10 full-sibs/clones matched to resident genotype (no added diversity, pure density); (3) Control — sham handling of 10 residents.
• Sampling: Genotype 30 random individuals per mesocosm at each of G0–G5 (n = 30 × 6 × 18 = 3,240 genomes). Also census N_e proxies (sex ratio, variance in family size) and vital rates each generation.
• Outcomes: (a) census growth rate λ; (b) N_e from temporal drift variance; (c) V_A(1) from temporal autocovariance across ≥500 neutral markers; (d) fitness (survivorship × fecundity) of immigrant vs resident lineages.

Expected signature: Genetic-rescue mesocosms should show immigrant alleles rising, elevated V_A(1) via positive temporal autocovariance, and higher λ; demographic augmentation should elevate λ transiently but leave V_A(1) and N_e unchanged. Control mesocosms calibrate drift-only decline in heterozygosity expected at N_e ≈ N = 100.

Analysis Pipeline

Compute per-locus Δp_t = p_t+1 − p_t; estimate N_e = 1/(2·Var(Δp)) after correcting sampling noise for n = 30; regress lag-1 covariance of Δp against LD-to-selected-marker distance to extract V_A(1) using the windowed genome-wide extension Temporal autocovariance code and method-of-moments estimator. Test treatment effects with mixed models (treatment fixed, mesocosm random, generation repeated).

Critical Limitations & Blindspots

(1) The autocovariance estimator assumes a closed, constant-size population with additive fitness effects and no migration Stated estimator limitations — the immigration pulse violates the closed-population assumption and must be modeled explicitly. (2) n = 30 genotypes per mesocosm gives high sampling variance on N_e; allele-frequency estimates will be noisy for rare immigrant alleles (expected frequency 10/110 ≈ 0.09). (3) Six generations may be too short to separate immigrant swamping from drift in small mesocosms. (4) Mesocosm conditions differ from the wild; the marten study shows that demographic benefits of augmentation often do not translate to field fitness Genotyping-error and immigration-data blindspots. (5) What would falsify the genetic-rescue hypothesis: immigrant allele loss with no V_A(1) elevation and no λ advantage relative to demographic treatment.