Explaining Paleoindian Settlement in the Intermountain West

background-color: black

.pull-lleft {
  float: left;
  width: 15%;
}

.pull-lright {
  float: right;
  width: 84%;
}

.pull-lright ~ * {
  clear: both;
}

</style>

]

Principals 
 - David Zeanah, Brian F. Codding, D. Craig Young, Robert Elston 
 
Field Directors 
 - Paul Algaier, Eric Martin, Kate Magargal, Ryan Bradshaw, Martijn Kuypers

Person responsible for current analysis
  - Kenneth B. Vernon

2021-04-21

]

---

## Archaeological Context

---

__Question__:  What explains Prearchaic settlement patterns in Grass Valley, Nevada?

]

---

__Question__:  What explains Prearchaic settlement patterns in Grass Valley, Nevada?

__Hypothesis__:  Individuals should occupy the most suitable habitats first, meaning those in which they can maximize their inclusive fitness.

]

---

__Question__:  What explains Prearchaic settlement patterns in Grass Valley, Nevada?

__Hypothesis__:  Individuals should occupy the most suitable habitats first, meaning those in which they can maximize their inclusive fitness.

__Observation__:  The rich, marshy environment surrounding the Pleistocene lake likely included the most profitable patches in the valley. Thus,

]

---

__Question__:  What explains Prearchaic settlement patterns in Grass Valley, Nevada?

__Hypothesis__:  Individuals should occupy the most suitable habitats first, meaning those in which they can maximize their inclusive fitness.

__Observation__:  The rich, marshy environment surrounding the Pleistocene lake likely included the most profitable patches in the valley. Thus,

__Prediction__:  Prearchaic sites are expected to occur at greater densities in proximity to the Pleistocene lake.

]

---

## Prearchaic Site Density

---

# Spatial Thinning

---

# Spatial Thinning

---

# Spatial Thinning

---

# Regression adjustment

The __actual density__ of archaeological sites, `$\lambda(s)$`, is the outcome of scientific interest. Simple log-linear model:

$$ log\; \lambda(s) = \alpha + \beta x(s) $$

However, this is subject to __sampling bias__, `$b(s)$`. Also, a simple log-linear model:

$$ log\; b(s) = \gamma + \delta z(s) $$

Simple linear response means we can combine them into a joint model of the __observed density__:

$$ log\; \lambda(s)b(s) = (\alpha + \gamma) + \beta x(s) + \delta z(s) $$

---

## Density Model

$$ \lambda(s) = \text{Time Period} * (\text{Lake} + \text{Springs} + \text{Streams}) $$

---

## Observation Model

$$ b(s) = \text{Distance to Road} + \text{Deposition Age} $$
.pull-left[

]

]

---

## Results

---

## Results

---

__Question__:  What explains Prearchaic settlement patterns in Grass Valley, Nevada?

__Hypothesis__:  Individuals should occupy the most suitable habitats first, meaning those in which they can maximize their inclusive fitness.

__Observation__:  The rich, marshy environment surrounding the Pleistocene lake likely included the most profitable patches in the valley. Thus,

__Prediction__:  Prearchaic sites are expected to occur at greater densities in proximity to the Pleistocene lake.

]

---

## Lingering Issues

__Environmental homogeneity__.

- Hard to tease out environmental drivers of settlement.

__Sampling bias__, again.

- This analysis accounts for only _surface_ sampling bias, and that only partially.
- Heaps of sampling bias likely still occur in intensely surveyed areas.

__Equifinality__.

-  The same observed density, `$\lambda(s)b(s)$`, could result from 
  - different combinations of `$\alpha$` and `$\gamma$`, and
  - different combinations of `$x(s)$` and `$z(s)$` when those are highly correlated.

---

## Conclusion

Need to...

* expand __geographic range__.

* invest in more __targeted inventory__, and

* incorporate __research history__ as prior.

---

## Acknowledgments

Special thanks to

* The University of Utah Archaeological Center Lab Group
* Kristen Hawkes and James F. O'Connell
* Kurt Wilson and Kasey Cole
* Peter Yaworsky, Weston McCool, and Kate Magargal
* Jon Wilker and the Gund Ranch Crew

]

This material is based upon work supported by the National Science Foundation under Grant No. (BCS-1632521, -1632522, -1632526).

]