Seasonality Revisions: Data + Modeling

[acannistra]

WIP

• Generate annual mean temperature for each isolate location
• Generate MHW-period mean temperatures for each MHW for each isolate location
• Re-compute MHW data containing new temperature metrics
• Compute "seasonal departure" metric as mhw_period_sst_mean – annual_sst_mean
• Modify models to replace season factor with "seasonal departure" metric
• Generate result figures for all model types

[acannistra]

Initial results with "seasonal departure" metric:

formula_abslat_re <- performance_diff_unscaled_mean ~ (abslat_scaled + seasonal_departure)^2 + (1|isolate)

Fit (summary):

Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula: formula_abslat_re
   Data: mhwPerformance

REML criterion at convergence: -14656.7

Scaled residuals: 
     Min       1Q   Median       3Q      Max 
-17.1513  -0.2356   0.0321   0.3195   7.3419 

Random effects:
 Groups   Name        Variance Std.Dev.
 isolate  (Intercept) 0.008355 0.0914  
 Residual             0.011151 0.1056  
Number of obs: 9170, groups:  isolate, 130

Fixed effects:
                                   Estimate Std. Error         df t value
(Intercept)                       2.046e-02  8.104e-03  1.280e+02   2.524
abslat_scaled                     2.689e-02  8.149e-03  1.282e+02   3.300
seasonal_departure               -7.944e-02  1.186e-03  9.050e+03 -66.976
abslat_scaled:seasonal_departure  2.812e-02  1.916e-03  9.046e+03  14.675
                                 Pr(>|t|)    
(Intercept)                       0.01282 *  
abslat_scaled                     0.00125 ** 
seasonal_departure                < 2e-16 ***
abslat_scaled:seasonal_departure  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Correlation of Fixed Effects:
            (Intr) abslt_ ssnl_d
abslat_scld  0.049              
sesnl_dprtr  0.001 -0.012       
abslt_scl:_ -0.005  0.025 -0.326

Predicted Relationships: