Effectiveness of assisted reproductive technologies for improved cattle and swine production in Rwanda

Abstract

Livestock production in Rwanda faces persistent genetic and productivity constraints under smallholder-dominated systems, where traditional breeding limits the speed and consistency of genetic improvement in cattle and swine. Assisted Reproductive Technologies (ARTs), including artificial insemination (AI), embryo transfer (ET), and multiple ovulation and embryo transfer (MOET), are increasingly promoted to accelerate genetic gain. However, their performance under field conditions is shaped by breed composition, physiological status, welfare standards, technical execution, and delivery logistics. This thesis synthesizes six empirical studies that quantify ART efficiency and identify biological and operational determinants of reproductive success in Rwanda. Study I evaluated AI outcomes in 351 cows and heifers under natural versus synchronized estrus. Pregnancy rate at 70 days post-AI was significantly higher under natural estrus (62.8%) than synchronized estrus (29.4%) (P < 0.01). In multivariable models, crossbred advantage was evident under natural estrus: Cross-Friesian (aOR = 3.88, 95% CI: 1.34–11.24; p = 0.01) and Cross-Jersey (aOR = 12.57, 95% CI: 2.83–55.84; p < 0.01) showed higher pregnancy odds relative to Ankole. Conversely, under synchronization, Cross-Friesian had reduced odds (aOR = 0.28, 95% CI: 0.12–0.66; p < 0.01). Timing was critical: insemination >18 hours after estrus detection markedly reduced pregnancy under natural estrus (aOR = 0.18, 95% CI: 0.04– 0.78; p = 0.02), and silent estrus reduced pregnancy under synchronization (aOR = 0.19, 95% CI: 0.06–0.61; p = 0.01). These findings indicate that estrus expression quality, breed compatibility, and precise timing outweigh protocol availability in smallholder contexts. Study II assessed ET success in 127 recipient cows, defining pregnancy at Day 52 post-transfer. Pregnancy was higher on public stations than smallholder farms in univariable analysis (44.8% vs 26.7%; p = 0.03), although this effect attenuated after adjustment (aOR = 1.78; p = 0.14). Embryo type emerged as a dominant determinant: fresh embryos achieved higher pregnancy than frozen embryos (54.3% vs 25.9%) and remained independently associated with success (aOR = 2.94, 95% CI: 1.32–6.52; p = 0.01). Recipient body condition also influenced outcomes; BCS 4 increased pregnancy odds compared with BCS 3 (aOR = 3.36, 95% CI: 1.02–11.07; p = 0.05). These results confirm ET feasibility but emphasize the importance of embryo quality and recipient physiological readiness. Study III examined welfare determinants of ET readiness and pregnancy among 208 synchronized recipients. Clear heat signs were observed at 89.9%, recipient suitability n(functional corpus luteum) was 61%, and pregnancy among suitable recipients was 36.2%. Welfare practices significantly influenced suitability and pregnancy in univariable analysis (P <0.01). In adjusted models, clean barn hygiene remained independently associated with pregnancy (aOR = 2.41, 95% CI: 1.01–5.76; p = 0.05), while ad-libitum feeding (aOR = 2.08) and water provision (aOR = 1.92) showed positive but attenuated effects. These findings position welfare as a mechanistic contributor to endocrine stability, uterine receptivity, and embryo survival. Study IV quantified super-ovulatory response and embryo yield in MOET programs using count data regression. Station, FSH type, and donor body weight significantly influenced outcomes. Kinigi outperformed Songa for CL (IRR = 2.31; p = 0.01), total structures (IRR = 2.08; p <0.01), and viable embryos (IRR = 2.42; p < 0.01). Stimufol increased CL (IRR = 1.94), total structures (IRR = 3.64), and viable embryos (IRR = 3.02) relative to Plus et (all p < 0.01). Donors weighing 400–500 kg demonstrated higher CL (IRR = 1.76; p = 0.01) and viable embryos (IRR= 1.84; p = 0.01) than >600 kg donors, indicating that metabolic balance, rather than maximal size, optimizes ovarian response. MOET efficiency thus depends on harmonizing endocrine stimulation with donor physiology and technical capacity. Study V compared swine natural service and AI in 583 sows. Natural mating modestly increased litter size (β = 0.75; p = 0.02) and weaned piglets (β = 0.69; p = 0.03). However, farm effects were substantially stronger; Station D increased litter size (β = 3.41; p < 0.01) and weaned piglets (β = 3.78; p < 0.01) relative to smallholders. These results demonstrate that management environment and herd-level practices outweigh service modality as determinants of productivity. Study VI evaluated drone facilitated semen delivery in pig AI systems. Conception was significantly associated with farming experience (1–3 years: aOR = 3.42; 5–8 years: aOR = 4.91) and AI experience ≥3 times (aOR = 5.84; p = 0.014), whereas transport duration and distance were not significant. Negative binomial models showed that university education increased litter size (IRR = 1.48), born alive (IRR = 1.55), and weaned piglets (IRR = 1.63) (all p < 0.01). Artificial insemination experience>3 times similarly improved litter and survival outcomes. These findings indicate that human capital and technical competence, rather than logistics alone, drive realized performance. Overall, ARTs can substantially enhance reproductive efficiency in Rwanda. However, success depends on breed-specific AI strategies, embryo and recipient quality in ET, welfare and physiological readiness, optimized hormonal protocols in MOET, and strong farm-level management supported by skilled operators and adaptive delivery systems.