Ines had set the kettle going at the small kitchenette down the hall and forgotten about it twice already. The third time she heard it click off she left it, because the query had returned.
Eleven seconds. She watched the timestamp on the panel and then watched it again, as if it might revise itself upward into something she recognized. In Lyon, on her own institution's compute, the same query — overnight body temperature distributions from biologger-tagged Marmota flaviventris in years of contrasting snowpack, cross-referenced against the vet-school pathogen panels, against the western pika range-contraction layer, against her own group's hantavirus surveillance from the Cantabrian and Carpathian transects — would not have been a query. It would have been a small project. Three years ago when she had first sketched what she would want, in the margin of a referee report she was procrastinating on, it had been twenty minutes of assembly and then a wait. Now it was a sentence she typed in a window. The synthesis interface had asked her, politely, whether she wanted to include the snowpack-timing covariates or treat them as a stratifying factor. She had said stratify. It had done the rest.
She scrolled. The colony-stratified output came back with the high colonies — River, Picnic — showing a pattern in the late-winter torpor bout structure during the low-snowpack years that she had not expected and did not yet trust. The pathogen panels came back thin in exactly the years she would have wanted them thickest. The pika layer had been updated last week; someone in Logan had pushed a revision. Her own European data sat alongside the basin's like a sibling that had grown up in a different house.
She made a note. Ask M. about the 2034–2036 panel gap — staffing? funding? deliberate? The synthesis lead was M. Halversen, whom she would meet tomorrow at eight. She had been briefed: Halversen had been at Gothic for eleven years, came up through entomology, ran the data team's synthesis side, was famously short with people who pretended to understand the marmot record without having sat with it.
Ines did not pretend. She had read the protocols on the plane.
Outside, the front coming over the ridge had darkened the upstairs office to the point where she reached for the desk lamp and then remembered the desk lamp had been replaced sometime in the spring with something that adjusted itself. It had already adjusted. She had not noticed.
Her phone buzzed: Salomé, her PhD student, from Lyon. Connection patchy on your end? Tower icon is red here.
Yes — front coming in. Try in ten?
She used the ten minutes to reread her note from the morning. She had been thinking about the question wrong. The question she had come with — whether reduced snowpack in continental mountain systems had already shifted the conditions under which hantaviruses establish in altered alpine mammal communities — was a population-level question dressed up as a mechanism question. The basin's data let her undress it. Individual marmots. Individual winters. Individual torpor bouts measured against individual fat-reserve estimates against pathogen exposure on the same animals across the same years. Her institution had the biostatistics. Her institution did not have Marmot 4719, born 2031, recaptured every year through 2041, biologged for three of those winters, sampled twice for the vet-school panel.
She had spent her second day at RMBL365 just reading 4719's file. Halversen had set it aside for her without being asked.
The kettle had cooled. She made tea anyway. The kitchenette had a window onto the parking lot where two technicians were loading a truck with what looked like soil-moisture nodes for redeployment. The taller one — Esme, she thought; they had been introduced at the Tuesday meeting — was arguing with the shorter one about whether to drive up to the Avery plots tonight or wait out the front. The shorter one wanted to drive. Esme wanted to wait. They were both right; the front was small but the road past Schofield had been bad since the June rains. Ines had walked part of it on Sunday with the partnership coordinator, who had pointed out where the willows had moved downslope in the last five years and where they had not. The coordinator's daughter, twelve, had been with them, narrating the entire walk in a register that suggested she had heard this conversation among adults her whole life and had developed firm views.
Back at her desk, Ines drafted a paragraph for the Lyon call.
The opposing-trends question is tractable here in a way it isn't anywhere else. Summer survival and winter survival behave as separate processes with separate climate drivers, and the basin's records carry enough individuals across enough colonies to estimate them jointly. What I want from you and Marek is the hierarchical model we sketched in February, but with the colony level as the elevation-by-snowpack interaction we couldn't parameterize in Europe because we don't have the colonies. Here we do. I will send you the data extract tonight — Halversen has cleared the share agreement, the synthesis interface will let me push a derived dataset to our group without moving the underlying records.
She paused on that last clause. Three years ago she would have had to argue for a data-use agreement that took six months. The cross-mountain residency had walked her past all of it. The agreement she had signed in June ran to four pages and committed her, among other things, to crediting the basin's records "as the substrate on which the synthesis was developed," which she had taken as fair and slightly unusual phrasing until Halversen's deputy explained at lunch that they had argued the language out with a working group of forty institutions over two years and this was what had survived.
The phone buzzed again. Salomé, video this time. The image came through grainy, froze, recovered.
"I can see you," Salomé said. "Mostly. Is that snow behind you?"
"Graupel. The front." Ines turned the laptop so Salomé could see the window. The light over the ridge had gone the color of wet slate. "It will pass in an hour."
"You look — you look like you have not slept."
"I slept. I have been reading one marmot's file for two days."
"Just one?"
"Born 2031. Female. Three winters of biologger data. The torpor structure in her low-snowpack winter is — it is not what the controlled-setting literature predicts. The bouts are shorter and the inter-bout arousals are warmer. She survived. Her two sisters did not."
Salomé was quiet. Then: "Is that the question, or is that a story?"
"It is one animal. It is a story. But there are forty-three females in the high colonies with comparable records and the question is whether the story holds across them. That is what I want to run before tomorrow."
"And the hantavirus side?"
"Thin. The pathogen panels are thin in the years I want them thickest. I have a question for Halversen about that. Whether it was a gap they will close, or a gap that is structural."
"Structural how?"
Ines hesitated. "They made choices, fifteen years ago, about what to build out and what to leave. The veterinary collaboration was one of the things they built. But not at the density I want. I think they would say — I think they would say that they chose to build the records and the tools and let people like us come with our own pathogen questions. Which is what I am doing."
"So you are the structural answer."
"I am one of them. There are eleven other groups using this interface this week. The Norwegian lemming people. Someone from Sapporo on Japanese serow. The pika consortium. We are all using the marmot records as a reference dataset because no one else has anything that goes back this far on individuals."
Salomé laughed. "You sound like you are going to write a grant."
"I am going to write three. Send me the February sketch. I want to push the extract tonight before the tower drops again."
"Tower?"
"The cell tower at Gothic. It has been intermittent. I am told this is unrelated to the weather, which I find difficult to accept."
"Sleep, Ines."
"After I push the extract."
She closed the call. The graupel had reached the window. She could hear it on the metal roof above the office, a tight dry sound, not quite hail.
She turned back to the interface and began to write the query that would generate the derived dataset for Lyon. She wanted it shaped so that Salomé and Marek could see immediately what the colony-by-snowpack interaction looked like for the high colonies, and so that the question for Halversen tomorrow — the question she could feel the shape of but not yet form — would be there in the data when she sat down at eight, in the form of a discrepancy she could point to. She thought it had to do with whether the warmer inter-bout arousals were an energetic cost the animal could carry indefinitely, or a debt that came due in the third or fourth such winter. The basin had not yet had a fourth such winter in sequence. It would, possibly soon.
The kettle clicked on again. She had not touched it.
She kept typing.