SRS still accounts for about one in ten sea-phase salmon deaths in Chile, a share that has held between 8 and 11.5% since 2018. We analysed 389 indexed SRS documents from 1989 onward and set them against Sernapesca mortality records. Research output nearly tripled in a decade and shifted toward control, especially antibiotic alternatives. Since 2023, though, 79% of studies were lab-based and only 10% used farm data. This piece covers where that gap sits, what the few farm-level studies show, and how records farms already keep could become comparable evidence.
Ahead of the SRS 2026 International Conference (Puerto Varas, Chile, 9–11 November 2026, organised by INCAR²), we ran a bibliometric analysis of the scientific literature on SRS in salmonids, 389 indexed documents from 1989 to today, and set it against Sernapesca's mortality records. Here is what came out of it.
In 1989, an outbreak in coho salmon farmed at sea near Puerto Montt killed around a million and a half fish. Farmers had been seeing the disease in coho since 1981. The 1989 outbreak was the one that let researchers isolate the agent (Fryer et al., 1990) and show that it caused the disease (Cvitanich et al., 1991). Today we know it as Piscirickettsia salmonis, and the disease as SRS or piscirickettsiosis. It shows up almost entirely in the seawater phase, usually a few weeks after transfer to sea.
Thirty-seven years on, about one in every ten salmon that die at sea in Chile dies of SRS. That share was 21 % in 2015 and, after two unusual years, has stayed between 8 and 11.5 % since 2018. In Atlantic salmon it is close to 12 %, and in rainbow trout it is higher, around 16 %. The disease is everywhere. In Los Lagos, fewer than 2 % of production cycles reached harvest without SRS in 2015–2016 (Bravo et al., 2020). In Aysén, in a typical week, more than half of the active farms have it (Lam et al., 2025). It is also still the main reason Chilean salmon farming uses antibiotics.
2026 looks somewhat better so far: 9.8 % of sea mortality in weeks 1–39, against 12.4 % for the same weeks of 2025. It's a good sign, but one season doesn't break a ten-year plateau.
SRS research almost tripled in ten years, from 8.4 documents a year in 2011–15 to 23.8 in 2021–25. Among the diseases of Chilean salmon, SRS is the one with the most research per point of mortality it causes, nearly twice any other. It is also research done from Chile: 78 % of the documents have at least one Chilean author.
What has changed is the question. For the first twenty years, almost half of the work studied the pathogen itself. Now most of it asks how to control it. Antibiotic alternatives (functional diets, plant extracts, peptides) went from 7.5 % to 22.5 % of the documents between 2018–22 and the period from 2023 onwards, and antimicrobials went from 10.5 % to 18 %.
This is where the research and the industry's problem drift apart. Since 2023, 79 % of SRS research has been done in the lab: in vitro work, in silico work, or challenge trials in experimental tanks. Only 10 % uses data from farms, the lowest share in the series.
None of this means lab work has failed. It is how we know what we know about the pathogen and how candidate tools get developed. And the lack of published field evidence doesn't mean companies have no data or experience; much of what the industry knows stays in internal records. The problem is that it rarely becomes shared, comparable evidence.
The gap isn't new, either. In 2003 a review was already asking for field trials to follow promising lab results (Fryer & Hedrick, 2003). In 2014 another found that most commercial vaccines had no apparent effect on field mortality (Rozas & Enríquez, 2014). The most recent review, by Marco Rozas-Serri (2026), calls it a "translational gap": we know more and more about the pathogen, and that knowledge doesn't turn into control at the farm.
The few studies built on production data tell a consistent story: SRS depends as much on management, timing and the neighbourhood as on any single product.
The neighbourhood matters. In Los Lagos, the strongest predictor of a farm's SRS prevalence was how many infected farms sat upstream of it, measured with official weekly records and an ocean-current model. Exposed farms came close to 100 % prevalence by week 25; isolated ones reached about 10 % by the end of the cycle (Bravo et al., 2020).
The time from signal to decision matters. In 571 Atlantic salmon cycles from Sernapesca's surveillance programme, cycles where antibiotic treatment started at the first signs of SRS mortality reached the official outbreak alert later than those that started once mortality was already rising (Gaete-Carrasco et al., 2026). This is a retrospective association and should be read carefully: fast, aggressive outbreaks leave less time to react, so part of the difference may come from the outbreak itself and not from the decision. The authors are explicit that this is not a case for using more antibiotics. It is a case for detecting and deciding in time. That window is often lost in operations: days can pass between diagnosis and the arrival of medicated feed, and by then part of the population is already eating less (Lie et al., 2025). In that same experimental tank trial, feed intake fell several days before the first death, a signal every farm already records.
Vaccination explains less of the difference than expected. Across more than 5,000 cycles from seven companies, some vaccination programmes (given in freshwater, before transfer to sea) did better than others. But most of the variation between cycles sat at the company level, in management, genetics or feed that the data didn't capture (Happold et al., 2020). There are also clues about why. Antibody levels peak 600–800 degree-days after vaccination and fall back to baseline around 1,800–1,900, while harvest comes at 3,000–6,000. Antibodies are not the same as clinical protection, but none of the labels of 28 registered vaccines reviewed says how long protection lasts (Valenzuela-Aviles et al., 2022). In the manufacturer's own lab trials, the live vaccine protected much less when fish were vaccinated in cold freshwater (Olsen et al., 2024).
Sea lice are part of the SRS problem. In an experimental tank trial, co-infection with Caligus cut the survival of vaccinated fish from 43 % to 5 % (Figueroa et al., 2017). The size of that effect won't carry over directly to a sea cage, but the direction is clear: a vaccine tested without lice promises more than it delivers on a farm with lice.
Health managers prioritise the basics. Nine fish health managers, from companies that together produce about 59 % of Chile's Atlantic salmon, rated 67 SRS control measures. At the top they put the basics done well: clean nets, mortality removal and classification, smolt quality and Caligus control. At the bottom were vaccines and revaccination at sea (Bustos et al., 2026). These are experienced judgements, not measured efficacy, and the panel disagreed on a third of the measures. The authors themselves call for standard outcome metrics so that measures can be compared in the field.
The visible cost is antibiotics. Ninety-two percent of florfenicol treatments in Chile are for SRS, and the median dose actually used is about 1.7 times the reference dose (Arriagada et al., 2024). The loss estimates that circulate (USD 450–700 million a year) come from reviews and the press; we found no published study that estimates the cost of SRS with a stated method.
Many of the questions the literature leaves open can be answered with data that farms already produce: daily mortality by cause, treatments, vaccination, feeding, sea lice counts, temperature and oxygen. The data is there. What's missing is putting it together and reading it in context.
At Manolin we think about this in three layers:
That opens up concrete questions:
Some of these questions can be explored today wherever records are consistent. Others need data that is rarely captured well, such as the product, date and water temperature of each vaccination, or the outbreak history of neighbouring farms. And all of them depend on permission to use the data, on standardisation across companies and on record quality. Cause of death is mostly assigned from clinical signs, and in the largest field study of vaccines, more than half of the cycles had to be excluded because their vaccination records were incomplete (Happold et al., 2020). Better records are already better evidence.
We'll be at the SRS 2026 International Conference. If you'd like to see the full analysis or talk through these questions with your own operation's data in mind, let's have a coffee during the conference.
Data: bibliometric analysis of 389 documents indexed in Scopus, 1989–2027 (search run on 30 October 2026; 2027 records are peer-reviewed articles in press; preprints are not counted). Mortality: Sernapesca (SAIP), deaths by cause as a % of total sea-phase mortality, 2015–2026 (2026 through week 39).