Table 5 GAM models of net primary production (NPP) for the oligohaline (OH) salinity zone.

From: Seasonal to Inter-Annual Variability of Primary Production in Chesapeake Bay: Prospects to Reverse Eutrophication and Change Trophic Classification

gam1.pp.net.oh <- gam(log_PP_net ~ s(log_Pbopt_net) + s(log_Euchl) + s(Kpar) + s(Temp) + s(Salin) + s(Month, bs = “cc”, k = 8) + s(Seq_year) + Seq_year + Season + s(LOG_SRF) + s(LOG_SUM) + s(TN_LDG) + s(NO23_LDG), data = PP_OH)a,b

gam2.pp.net.oh <- gam(log_PP_net ~ s(log_Pbopt_net) + s(log_Euchl) + s(Kpar) + s(Temp) + s(Salin) + s(Month, bs = “cc”, k = 8) + Season + s(LOG_SRF) + s(LOG_SUM) + s(TN_LDG) + s(NO23_LDG), data = PP_OH)c

gam3.pp.net.oh <- gam(log_PP_net ~ s(log_Pbopt_net) + s(log_Chl) + s(Kpar) + s(Temp) + s(Salin) + s(Month, bs = “cc”, k = 8) + Season + s(LOG_SRF) + s(LOG_SUM) + s(TN_LDG) + s(NO23_LDG), data = PP_OH)d

gam4.pp.net.oh <- gam(log_PP_net ~ s(log_Pbopt_net) + s(log_Chl) + s(Temp) + s(Salin) + s(Month, bs = “cc”, k = 8) + Season + s(LOG_SRF) + s(LOG_SUM) + s(TN_LDG) + s(NO23_LDG), data = PP_OH)e

gam5.pp.net.oh <- gam(log_PP_net ~ s(log_Pbopt_net) + s(log_Chl) + s(Temp) + s(Salin) + s(Month, bs = “cc”, k = 8) + s(LOG_SRF) + s(LOG_SUM) + s(TN_LDG) + s(NO23_LDG), data = PP_OH)d,e

  1. agam models of NPP for MH and PH salinity zones had the same structures as these models, using input data PP_MH and PP_PH; models based on the all calibration data (1982 to 2004) used input data PP_ALL and added a categorical term for Salzone (OH, MH, PH);
  2. bgam models to predict GPP had the same structure as those for NPP with log_Pbopt_gross substituted for log_Pbopt_net;
  3. bgam2 models applied to water-quality data from 1985 to 2015 were used to predict NPP from data files WQ_OH, WQ_MH, and WQ_PH; the explanatory variable Seq_year was omitted from these models to avoid extending trends in calibration data (1982 to 2004) to years outside that time frame;
  4. cSubstituted log_Euchl with log_Chl in gam models to predict NPP to test models using input data that lacked euphotic-layer chl-a;
  5. dAnnual TN and NO2 + NO3 loadings were replaced with monthly loadings in gam models to predict NPP for historical data, based on the lack of data on Kpar or monthly loadings for the 1960s and 1970s;
  6. eSeason was omitted as a categorical variable in gam models to predict NPP due to small sample sizes for historical data.
Back to article page