Shared range predicted which Illinois doe groups came close

STEP 1 · Between groups

Same-group does were set aside on purpose

The telemetry pool held 69 female deer from five study areas, with collar years running from 2002 to 2014. Koen and colleagues set aside any collared doe that belonged to the same social group as another collared doe. Earlier papers had flagged those groups from correlated movement and consistent proximity. Every link that remained runs between neighboring groups. The analysis does not describe meetings inside one family group. Relatedness of the collared does was unknown.

Pair-by-pair models used the sets with all three seasons. Carbondale from 2002 to 2006 had 24 does. Lake Shelbyville from 2007 to 2009 had 12 does. Our addition of 24 and 12 is 36. The abstract names a neighboring-group sample of 36. Carbondale seasonal networks ranged from 3 to 11 collared does and averaged 5.9. Lake Shelbyville networks ranged from 2 to 10 and averaged 5.4. Locations were kept every 2 hours. Overlap was a seasonal map of shared ground and did not require two does there in the same hour. Gestation ran from 1 January through 14 May. Fawning ran from 15 May through 31 August. The rut window ran from 1 September through 31 December and is not a legal season. The authors kept pairs with enough simultaneous fixes and overlapping ranges, and they said the filtered sample was not truly random.

Stationary collars under closed canopy had a median position error of 29 feet (8.8 m). That error comes from earlier work on stationary collars and sits inside the 82-foot association rule. It does not describe moving deer or open ground.

STEP 2 · Shared ground

Overlap led most of the seasonal models

Does that shared more space tended to have higher association rates across these local models. Overlap was the best overall predictor of sociality among the groups. It had the greatest influence during gestation at both Carbondale and Lake Shelbyville. During fawning at Carbondale, forest connectivity and overlap together best predicted sociality. The Lake Shelbyville fawning models did not detect a land-cover effect or an overlap effect. That miss leaves the question open. Overlap was not the lead predictor in the Lake Shelbyville rut models.

STEP 3 · Fawning woods

Narrow forest corridors went with less sociality at Carbondale

Association rates tended to be lower during fawning, when adult does tend to be solitary. At Carbondale the cover mixed forest and grassland. Does in highly connected forest tended to be less social with neighbors.

The circuit model scored forest connectivity highest in narrow corridors of woods surrounded by open ground. The authors called those corridors pinch points. Pixels inside wider forest tracts had a lower modeled probability of movement. The does appeared more social where the forest tracts were wider. That score is modeled current from a land-cover map.

Our read is that a narrow wooded corridor during fawning is a weak reason to expect neighboring doe groups to stand close. The reading belongs to the Carbondale fawning sample. A connected cover map alone is not a record of which does stood near each other.

STEP 4 · Soft tests

Crop signals were weak in the rut and across seven networks

During the rut window at Lake Shelbyville, the best predictors of high sociality were a high amount of agriculture and high connectivity of agriculture. The 95 percent confidence intervals on those coefficients excluded zero. The model statistics beat 95 percent of randomized values. The null model stayed competitive, so the authors call the landscape evidence weak. Our read in crop country for that window is to keep weighing shared range and to let your own notes test any crop-field pattern.

Across seven population networks in early gestation, more agriculture and less forest tended to go with higher closeness. The fits were an R2 of 0.40 for agriculture and 0.39 for forest. Connected forest-agriculture edge tended the same way at an R2 of 0.38. The small number of networks kept the fits from statistical significance. A missed cutoff leaves a real landscape pattern and a chance pattern both possible.

The authors treat that pattern as a hint that infectious disease could spread faster where agriculture is abundant, forest is scarce, and forest-agriculture edge is connected. No disease was measured in these deer. Range overlap can explain the association patterns in this sample. It cannot quantify transmission. It cannot justify a disease-control treatment. When the paper was written, the authors said chronic wasting disease had been detected in Wisconsin and northern Illinois and not yet in the central and southern Illinois study areas. They wrote that their study populations were not impacted. That statement is from the time of the paper. Check your state agency for current detections.

Land cover came from 1999 and 2000 imagery. Lake Shelbyville was mostly corn and soybean, while Crab Orchard and Touch of Nature were mostly forest. Some collars were still on deer in 2014, years after that imagery. Networks with few nodes still appeared to have higher closeness after the authors adjusted for node count. They leave open whether node count still shapes the landscape relationship.

STEP 5 · Field notes

Sightings record proximity, not this network

An optional log uses the outings you already make. Note which bedding and feeding patches the does use. On outings when more than one of those patches is in view, note whether the deer were near one another. Unmarked does usually cannot be sorted into known unrelated social groups from a few photos. The log records observed proximity and unknown membership. It does not rebuild the group-to-group network in this paper. It cannot prove the modeled connectivity score or a cause. It cannot show what the whole herd does.

Our expectation from the overlap pattern is that does seen using the same bedding and feeding patches will be seen close together more often than does from patches with little shared use. Unmarked deer make this a rough check, not a measure of overlap. That expectation is ours to test on your ground. Daylight sightings are a rough stand-in for fixes taken every 2 hours around the clock. Adult does tend to be solitary in the fawning window these authors used, so groups are harder to watch then. A thin tally stays a hint.

The ranking can only be tested on your ground where you can recognize the same does over time, for example by distinctive marks. If does that rarely use the same patches show up close as often as does that clearly share patches, treat the overlap-first ranking as weaker there. Without recognizable deer, the log stays a hint and cannot overturn either reading. Either change applies to your ground only.

The collars cannot settle buck movement or herd size. They cannot settle hunting success or whether any disease is present. Citing earlier work, the paper notes that does in the same social group associate more than neighbors do. That within-group pattern was not measured here, because same-group does were removed on purpose.