Genomic epidemiology of the 2017-2023 outbreak of Mycoplasma bovis sequence type ST21 in New Zealand

This study utilizes genomic epidemiology and phylodynamic modeling to track the 2017–2023 *Mycoplasma bovis* outbreak in New Zealand, demonstrating how movement restrictions and culling successfully reduced transmission and eliminated major lineages by 2020, while highlighting the critical role of integrated genomic surveillance in guiding eradication efforts.

The Gist

Imagine New Zealand's cattle industry as a massive, bustling city of cows. In 2017, a sneaky, invisible burglar named Mycoplasma bovis broke in. This burglar doesn't just steal; it makes the cows sick with pneumonia, bad joints, and other nasty ailments. Because New Zealand had never seen this burglar before, the entire city was vulnerable.

The government decided they couldn't just lock the doors; they had to find every single burglar, kick them out, and burn the evidence. This was the "Eradication Program." But how do you catch a burglar you can't see? You need a super-powered detective team. In this case, the detectives were genomic epidemiologists—scientists who read the "DNA fingerprints" of the bacteria.

Here is the story of how they solved the case, told in simple terms:

1. The DNA Fingerprinting (The "Who's Who" List)

When the police (scientists) found the bacteria on a farm, they didn't just take a photo; they took a full DNA test. They did this for nearly 700 cows on 126 different farms.

Think of the bacteria like a family tree. By comparing the tiny differences in their DNA, the scientists could draw a family tree showing who was related to whom.

• The Result: They found that the burglar likely arrived in New Zealand only once, around 2016 or early 2017. It wasn't a massive invasion from many different countries; it was one single entry that then split into three main "families" (or lineages) of bacteria.

2. The Speed of the Chase (The "Reff" Meter)

The scientists had a special speedometer called the Effective Reproduction Number (Reff).

• Before the crackdown: The speedometer read 2.5. This meant that for every one infected farm, it was infecting 2.5 new farms. The burglar was running wild.
• After the crackdown: The government put up roadblocks (movement restrictions) and removed the infected cows (culling). The speedometer dropped to 0.5. This means the burglar was now infecting less than one new farm for every one it was on. The fire was being put out.

3. The "Long Tail" Problem

By 2020, the scientists thought they had won. Two of the three bacterial "families" had disappeared completely. But the third family refused to die out.

This is where the story gets interesting. The scientists discovered a Giant Feedlot (a huge farm where cattle are fattened up for slaughter) in the South Island.

• The Metaphor: Imagine a giant, noisy party where the burglar is hiding. Everyone thought the feedlot was just a "trash can" where sick cows went to die. But the DNA evidence showed the feedlot was actually a super-spreader hub.
• The bacteria were living there for four years, hopping from the feedlot to nearby farms, and then hopping back. It was like a virus in a crowded subway station that never stopped moving.

4. The Final Blow

The scientists used their DNA maps to prove that this specific feedlot was the source of the remaining infections. They showed the government: "Look, all these other sick farms are connected to this one big farm."

Once the feedlot was finally cleared out (depopulated) in late 2022, the "long tail" of the infection finally stopped. The bacteria had nowhere left to hide.

5. The Big Picture: Why This Matters

This paper is a victory story for genomic surveillance.

• Old Way: You wait for a cow to get sick, then you guess where it came from.
• New Way: You read the bacteria's DNA like a GPS tracker. You can see exactly which farm infected which other farm, even if they never traded cows directly.

The Takeaway:
New Zealand successfully wiped out a major cattle disease that had plagued other countries for decades. They did it by combining old-school detective work (tracing cow movements) with high-tech DNA sleuthing. The DNA told them when the burglar arrived, how it spread, and where the last hiding spot was.

It's a perfect example of how reading the "code of life" can save an entire industry, turning a potential disaster into a success story of science and teamwork.

Technical Summary

1. Problem Statement

Mycoplasma bovis is a globally significant cattle pathogen causing pneumonia, mastitis, arthritis, and reproductive disorders. In July 2017, it was first detected in New Zealand (NZ), a country previously free of the pathogen. Given NZ's heavy reliance on dairy and beef exports, the detection posed a severe economic and welfare risk. The government launched an ambitious "Eradication Programme" involving surveillance, movement controls, and culling. However, the outbreak persisted longer than initially anticipated, particularly in the South Island, raising questions about transmission dynamics, the success of interventions, and the potential for multiple incursions. The challenge was to utilize genomic data to understand the outbreak's origin, track transmission pathways, and validate the effectiveness of the eradication strategy over a six-year period (2017–2023).

2. Methodology

The study employed a comprehensive phylodynamic approach, integrating whole-genome sequencing (WGS) with epidemiological data.

• Data Collection:
  • Samples: 1,026 isolates were sequenced from 697 cattle across 126 infected farms (representing 45% of all 282 confirmed infected farms).
  • Timeline: Samples collected from July 2017 to December 2023.
  • Sequencing: Illumina short-read sequencing for the majority of isolates; a reference genome (NCBI CP192245.1) was generated using PacBio long-read sequencing.
• Bioinformatics & Phylogenetics:
  • Assembly & Variant Calling: Reads were trimmed (Trimmomatic), assembled (SKESA/Nullarbor), and aligned to the reference genome. SNPs were identified using Snippy, and recombinant regions were removed using Gubbins.
  • Population Structure: Hierarchical clustering (rhierBAPS) was used to define clades.
  • Phylogenetic Modeling: Bayesian inference was performed using BEAST 2.7.8. Models included HKY substitution, strict/relaxed molecular clocks, and various tree priors (BICEPS, Birth-Death Skyline).
• Phylodynamic Modeling:
  • tMRCA Estimation: Used to estimate the Time of the Most Recent Common Ancestor to determine the likely time of incursion.
  • Effective Reproduction Number ($R_{eff}$) & Population Size ($N_e$): Estimated using the Birth-Death Skyline (BD-SKY) and BICEPS models to track transmission intensity and pathogen diversity over time.
  • Transmission Networks:
    • SCOTTI (Structured COalescent Transmission Tree Inference): Used to infer direct and indirect transmission probabilities between farms, accounting for the period of infectiousness.
    • MASCOT-GLM (Marginal Approximation of the Structured Coalescent): Used to model long-range dispersal (specifically between North and South Islands) and identify migration events.

3. Key Contributions

• Real-time Decision Support: This paper documents one of the few instances where genomic surveillance and phylodynamic modeling were used continuously throughout an active animal disease eradication program to inform policy (e.g., movement restrictions, depopulation timing).
• Resolution of the "Long Tail": The study provided the genetic evidence explaining why the outbreak persisted for years after the initial peak, identifying a specific large feedlot as a persistent reservoir.
• Differentiation of Incursions vs. Evolution: The analysis helped distinguish between a single introduction followed by rapid evolution versus multiple independent incursions.
• Quantification of Transmission Dynamics: It quantified the shift from long-range (inter-island) transmission to localized (within-farm and local spillover) transmission.

4. Key Results

• Origin and Incursion:
  • The estimated tMRCA for the farm-level dataset was April 2016 (95% HPD: Aug 2015–Oct 2016), suggesting a single incursion event roughly one year prior to the first detection in 2017.
  • The isolates formed a monophyletic clade (ST21), supporting a single introduction rather than multiple independent incursions, though rapid early divergence into three lineages (a, b, c) occurred.
• Impact of Interventions:
  • $R_{eff}$ Decline: The effective reproduction number dropped sharply from ~2.5 to <0.5 by mid-2018/2019, coinciding with the implementation of the eradication program (movement controls and culling).
  • Lineage Extinction: Two of the three initial lineages (b and c) went extinct by 2020.
• The "Long Tail" and the Feedlot:
  • From 2021 to 2023, the outbreak was confined to the Canterbury region (South Island).
  • SCOTTI and MASCOT analyses identified a large beef feedlot as the primary source of persistent infection. This farm was infected from August 2018 until its depopulation in December 2022.
  • Genomic data revealed the feedlot acted as a source (not just a sink) for local transmission, seeding multiple nearby farms.
• Inter-Island Transmission:
  • MASCOT-GLM models estimated multiple transmission events between the South and North Islands (median 12 South-to-North, 5 North-to-South) occurring primarily before 2019.
  • No inter-island migration events were detected after the end of 2019, confirming the success of movement controls in isolating the remaining infection to the South Island.
• Eradication Status:
  • Following the depopulation of the feedlot and surrounding farms, no new cases were detected. A separate detection of ST29 in 2022 was linked to imported semen and contained to a single farm, confirming the main ST21 outbreak was under control.

5. Significance

• Proof of Concept for Eradication: This study demonstrates the feasibility of eradicating M. bovis in a large, island-based cattle industry through a combination of aggressive culling, movement restrictions, and genomic surveillance.
• Role of Genomics in Outbreak Response: It highlights that genomic data is not just for retrospective analysis but is critical for real-time decision-making. It helped prioritize investigations (focusing on extant lineages rather than all historical farms) and challenged assumptions about transmission (e.g., identifying the feedlot as a source).
• Methodological Integration: The paper serves as a blueprint for integrating epidemiological tracing with phylodynamic modeling (SCOTTI, MASCOT) to resolve complex transmission networks where traditional tracing data is ambiguous or incomplete.
• Global Relevance: The findings offer valuable insights for other nations facing M. bovis incursions, particularly regarding the risks of persistent reservoirs in high-density farming operations (feedlots) and the importance of monitoring effective reproduction numbers to gauge program success.