EnvRtype is an R package for enviromics — the study of the envirome, the set of environmental conditions linked to the biological performance of living beings (Costa-Neto and Fritsche-Neto, 2021; Crossa et al., 2021). It collects worldwide daily weather and soil data, derives agro-meteorological parameters, builds environmental covariable matrices and relatedness kernels, mines environmental typologies, delineates soil zones with Gaussian mixture models, and runs a FAO-56 soil water balance for use in enviromics and genotype-by-environment (GxE) analyses.
This is a fully re-engineered successor to the original EnvRtype (allogamous/EnvRtype, 2021), extending it from a weather-typing toolbox into an end-to-end envirotyping, simulation and prediction framework.
📄 Original publication: Costa-Neto, G., Galli, G., Carvalho, H. F., Crossa, J., & Fritsche-Neto, R. (2021). EnvRtype: a software to interplay enviromics and quantitative genomics in agriculture. G3 Genes|Genomes|Genetics, 11(4), jkab040. Read the paper →
if (!require("devtools")) install.packages("devtools")
devtools::install_github("gcostaneto/EnvRtype")library(EnvRtype)The original EnvRtype (Costa-Neto et al., 2021, G3) focused on collecting NASA POWER weather data, computing environmental typologies and building environmental kernels for reaction-norm models. This version keeps that workflow and adds several new layers:
| Area | Original EnvRtype (2021) | EnvRtype (this version) |
|---|---|---|
| Weather data | get_weather() (NASA POWER, daily) |
get_weather() + hourly
(get_weather_hourly()) and resumable /
restartable downloads (get_weather_resumable(),
read_progress_log(), restart_from_log()) |
| Soil data | — | get_soil(), get_soil_resumable(),
soil_classification() (Gaussian-mixture soil zoning) |
| Other geodata | — | get_elevation(), get_bioclim(),
get_spatial(), get_AEZ() (with helpers
aez_legend(), aez_default_url()),
get_climate_scenario() |
| Processing | processWTH(),
param_temperature/radiation/atmospheric(),
summaryWTH() |
Same, plus a full FAO-56 water balance
(water_balance(), summary_water_balance()) and
phenology (env_phenology(),
phenology_templates(),
planting_window_table(), best_planting_date(),
show_phenology(), plot_planting_window()) |
| Characterisation | W_matrix(), env_typing() |
Adds T_matrix(), env_indices(),
env_expand(), a full PCA suite
(env_pca(), env_pca_biplot(),
env_pca_scree(), env_loading_curve(),
env_pc_associate()), correlation tools
(env_cor(), env_cor_heatmap()) and
coverage/target diagnostics (coverage_summary(),
env_target_importance()) |
| Risk & TPE | — | env_risk_profile(), env_copula(),
tpe_weights(), project_risk(),
env_target_importance() |
| Kernels | env_kernel(), get_kernel() |
Adds decompose_kernels(),
undecompose_kernels(), truncate_gxe_kernel()
and a soil kernel path in get_kernel() |
| Modelling | kernel_model() |
Adds kernel_cv(),
kernel_model_clustered(), kernel_model_mc(),
varcomp_summary(), environment clustering
(env_cluster(), cluster_environments()) |
| Untested environments | — | scan_untested_envs(), grid_scan(),
map_scan(), scan_spatial_table() |
| Simulation | — | Ground-truth simulator: sim_met(),
sim_met_C(), sim_W(),
sim_W_grid(), plus controllable-diversity data generators
sim_markers() (genetic diversity) and
sim_envirome() (environmental diversity), to validate every
layer against a known truth |
| License | MIT | GPL-3 (CRAN-ready) |
The package is organised in four data layers — acquisition, processing, characterisation and prediction — plus a simulation engine that generates ground-truth data to validate each layer. Each subsection lists the functions involved and the primary references used to build them.
Design follows the enviromics reaction-norm framework of Costa-Neto et al. (2021, G3) and its envirome-wide extension (Costa-Neto et al., 2023, G3).
flowchart TB
L1["<b>1 · ACQUIRE</b><br/>get_weather · get_soil · get_elevation<br/>get_bioclim · get_spatial · get_AEZ<br/>get_climate_scenario · + resumable/hourly variants"]
L2["<b>2 · PROCESS</b><br/>processWTH · summaryWTH · param_*<br/>water_balance · env_phenology"]
L3["<b>3 · CHARACTERISE</b><br/>W_matrix · T_matrix · env_indices · env_pca<br/>soil_classification · env_typing · env_risk_profile"]
L4["<b>4 · MODEL & SCAN</b><br/>env_kernel · get_kernel · kernel_model<br/>kernel_cv · scan_untested_envs · map_scan"]
SIM["<b>SIMULATE</b><br/>sim_markers · sim_envirome<br/>sim_met() → C_env → sim_W()<br/><i>ground truth for every layer</i>"]
L1 --> L2 --> L3 --> L4
SIM -.->|"validates"| L3
SIM -.->|"validates"| L4
classDef l1 fill:#1565c0,stroke:#0d47a1,color:#fff
classDef l2 fill:#1976d2,stroke:#0d47a1,color:#fff
classDef l3 fill:#2e7d32,stroke:#1b5e20,color:#fff
classDef l4 fill:#c62828,stroke:#b71c1c,color:#fff
classDef sim fill:#6a1b9a,stroke:#4a148c,color:#fff
class L1 l1
class L2 l2
class L3 l3
class L4 l4
class SIM simWeather and geodata come from NASA POWER via nasapower (Sparks, 2018), SoilGrids 2.0 (Poggio et al., 2021), WorldClim 2 (Fick & Hijmans, 2017), bioclimatic predictors (O’Donnell & Ignizio, 2012), CMIP6 / ScenarioMIP scenarios (Hawkins et al., 2013; O’Neill et al., 2016) and GAEZ v4 (FAO & IIASA, 2021). Agro-meteorological processing, the FAO-56 soil water balance and phenology follow Allen et al. (1998), Pereira et al. (2021), Doorenbos & Kassam (1979), Steduto et al. (2012), Borg & Grimes (1986), Forsythe et al. (1995), Zadoks et al. (1974) and Counce et al. (2000).
flowchart LR
GW["get_weather()"]
GWH["get_weather_hourly()"]
GWR["get_weather_resumable()"]
RPL["read_progress_log()"]
RFL["restart_from_log()"]
GS["get_soil()"]
GSR["get_soil_resumable()"]
GEL["get_elevation()"]
GBC["get_bioclim()"]
GSP["get_spatial()"]
GAEZ["get_AEZ()"]
GCS["get_climate_scenario()"]
PRAD["param_radiation()"]
PATM["param_atmospheric()"]
PTMP["param_temperature()"]
PWT["processWTH()"]
SWT["summaryWTH()"]
EPH["env_phenology()"]
PHT["phenology_templates()"]
PWD["planting_window_table()"]
BPD["best_planting_date()"]
SPH["show_phenology()"]
PPW["plot_planting_window()"]
WBAL["water_balance()"]
SWB["summary_water_balance()"]
GWR --> RPL --> RFL --> GW
GW --> PWT
GWH --> PWT
PRAD --> PWT
PTMP --> PWT
GEL --> PATM
PATM --> PWT
PWT --> SWT
PWT --> EPH
PHT --> EPH
EPH --> PWD
EPH --> SPH
PWD --> BPD
PWD --> PPW
PWT --> WBAL
GSR --> GS
GS --> WBAL
GEL --> WBAL
WBAL --> SWB
classDef collect fill:#1565c0,stroke:#0d47a1,color:#fff
classDef char fill:#2e7d32,stroke:#1b5e20,color:#fff
classDef wb fill:#0097a7,stroke:#006064,color:#fff
class GW,GWH,GWR,RPL,RFL,GS,GSR,GEL,GBC,GSP,GAEZ,GCS,PRAD,PATM,PTMP,PWT,SWT collect
class EPH,PHT,PWD,BPD,SPH,PPW char
class WBAL,SWB wbEnvironmental covariables, typologies and kernels follow Costa-Neto et al. (2021, G3); interval-resolved indices follow Della Coletta et al. (2023); breeding-zone typologies and target-population importance follow Costa-Neto et al. (2023, Agronomy Journal); copula reparameterisation follows Sklar (1959), Genest & Rivest (1993) and Salvadori et al. (2007); Gaussian-mixture soil zoning follows Fraley & Raftery (2002), Scrucca et al. (2016) and Egozcue et al. (2003).
flowchart LR
SWT["summaryWTH()"]
GS["get_soil()"]
WM["W_matrix()"]
ETYP["env_typing()"]
TM["T_matrix()"]
EEX["env_expand()"]
ECOP["env_copula()"]
ERP["env_risk_profile()"]
TPW["tpe_weights()"]
PRK["project_risk()"]
ETI["env_target_importance()"]
CVS["coverage_summary()"]
GCS["get_climate_scenario()"]
SCL["soil_classification()"]
EIX["env_indices()"]
EPC["env_pca()"]
EPS["env_pca_scree()"]
EPB["env_pca_biplot()"]
ELC["env_loading_curve()"]
ECOR["env_cor()"]
ECH["env_cor_heatmap()"]
EPA["env_pc_associate()"]
SWT --> WM
WM --> ETYP
ETYP --> TM
WM --> EEX
WM --> ECOP
WM --> ETI
ERP --> TPW
ERP --> PRK
GCS --> PRK
TPW --> CVS
GS --> SCL
EIX --> WM
EIX --> EPC
EIX --> ECOR
ECOR --> ECH
EPC --> EPS
EPC --> EPB
EPC --> ELC
EPC --> EPA
classDef char fill:#2e7d32,stroke:#1b5e20,color:#fff
classDef dis fill:#6a1b9a,stroke:#4a148c,color:#fff
classDef collect fill:#1565c0,stroke:#0d47a1,color:#fff
class WM,ETYP,TM,EEX,ECOP,ERP,TPW,PRK,ETI,CVS,SCL char
class EIX,EPC,EPS,EPB,ELC,ECOR,ECH,EPA dis
class SWT,GS,GCS collectReaction-norm and kernel models follow Costa-Neto et al. (2021, G3), the nonlinear / Gaussian kernels of Costa-Neto, Fritsche-Neto & Crossa (2021, Heredity) and the deep kernels of Cuevas et al. (2019); the GxE modelling engine follows the BGGE approach of Granato et al. (2018); scanning of untested environments follows the envirome-wide prediction of Costa-Neto et al. (2023, G3).
flowchart TB
EK["env_kernel()"]
GK["get_kernel()<br/>K_G · K_E · K_S"]
DK["decompose_kernels()"]
UDK["undecompose_kernels()"]
TGK["truncate_gxe_kernel()"]
KM["kernel_model()"]
KCV["kernel_cv()"]
KMC["kernel_model_clustered()"]
KMM["kernel_model_mc()"]
ECL["env_cluster()"]
CLE["cluster_environments()"]
VCS["varcomp_summary()"]
SUE["scan_untested_envs()"]
GSC["grid_scan()"]
MSC["map_scan()"]
SST["scan_spatial_table()"]
CVS["coverage_summary()"]
EK --> GK
GK --> DK
DK --> UDK
DK --> TGK
TGK --> KM
DK --> KM
KM --> KCV
KM --> VCS
KM --> SUE
ECL --> CLE
CLE --> KMC
KM --> KMM
SUE --> GSC
SUE --> MSC
SUE --> SST
SUE --> CVS
classDef pred fill:#c62828,stroke:#b71c1c,color:#fff
classDef rel fill:#ef6c00,stroke:#e65100,color:#fff
classDef char fill:#2e7d32,stroke:#1b5e20,color:#fff
class GK,DK,UDK,TGK,KM,KCV,KMC,KMM,VCS,SUE,GSC,MSC,SST,CVS pred
class EK rel
class ECL,CLE charThe simulator encodes a known environmental covariance and
reaction-norm structure to benchmark every layer, following the
envirome-wide prediction framework of Costa-Neto et al. (2023, G3) and
the kernel models of Costa-Neto et al. (2021, G3).
sim_markers() and sim_envirome() additionally
generate marker and envirome data with controllable genetic and
environmental diversity (MAF spectrum, population structure, linkage;
mega-environments, gradients and covariable redundancy), feeding
sim_met() and get_kernel() directly.
flowchart LR
SMK["sim_markers()<br/><i>genetic diversity</i>"]
K["genomic kinship K<br/><i>n × n lines</i>"]
SEV["sim_envirome()<br/><i>environmental diversity</i>"]
SMC["sim_met_C()<br/><i>q × q envs</i>"]
SM["sim_met()"]
CE["$C_env<br/><i>q × q correlation<br/>among environments</i>"]
PH["$data<br/>env · gid · rep · value"]
SW["sim_W()"]
SWG["sim_W_grid()"]
W["simulated W<br/><i>known % explained</i>"]
GK["get_kernel()"]
KM["kernel_model()"]
VCS["varcomp_summary()"]
CHK{{"compare to<br/>known truth"}}
SMK --> K
K --> SM
K --> GK
SEV --> SM
SEV --> GK
SMC --> SM
SM --> CE
SM --> PH
CE --> SW
SW --> W
SWG --> W
W --> GK
PH --> KM
GK --> KM
KM --> VCS
VCS --> CHK
SM -.->|"$truth"| CHK
classDef sim fill:#6a1b9a,stroke:#4a148c,color:#fff
classDef pred fill:#c62828,stroke:#b71c1c,color:#fff
classDef data fill:#455a64,stroke:#263238,color:#fff
classDef chk fill:#f9a825,stroke:#f57f17,color:#000
class SM,SW,SWG,SMC,SMK,SEV sim
class GK,KM,VCS pred
class K,CE,PH,W data
class CHK chkEnvRtype fits enviromics-enabled reaction-norm models in which phenotypes are regressed on genomic, environmental and GxE relatedness kernels. The current prediction workflow provides:
get_kernel()
assembles genomic (K_G), environmental (K_E) and soil (K_S) kernels;
env_kernel() builds Gaussian / linear environmental
kernels; decompose_kernels(),
undecompose_kernels() and
truncate_gxe_kernel() manage the GxE block structure.kernel_model() fits
the multi-kernel Bayesian reaction-norm model;
kernel_model_mc() runs multiple chains;
kernel_model_clustered() fits environment-clustered models
built with env_cluster() /
cluster_environments().kernel_cv() runs cross-validation (CV1 / CV2 / CV0 schemes)
and varcomp_summary() reports variance components and the
share of variance explained by each kernel.scan_untested_envs() predicts genotype performance in
environments with no phenotypes, while grid_scan(),
map_scan(), scan_spatial_table() and
coverage_summary() summarise the target population of
environments (TPE).These tools implement the reaction-norm and kernel methods of Costa-Neto et al. (2021, G3), the nonlinear / Gaussian kernels of Costa-Neto, Fritsche-Neto & Crossa (2021, Heredity), the deep kernels of Cuevas et al. (2019, G3), the BGGE GxE engine of Granato et al. (2018, G3), and the envirome-wide prediction of untested environments of Costa-Neto et al. (2023, G3). Full citations are listed under References for the new functions below.
EnvRtype began in 2020 and was first published as allogamous/EnvRtype in Costa-Neto, G., Galli, G., Carvalho, H. F., Crossa, J., & Fritsche-Neto, R. (2021). EnvRtype: a software to interplay enviromics and quantitative genomics in agriculture. G3: Genes|Genomes|Genetics, 11(4), jkab040. We gratefully acknowledge the original authors and the community of users whose feedback shaped this new version.
The methods and data sources underlying the new and extended functions are documented below, grouped by module.
env_copula)If you use EnvRtype in your research, please cite the original publication:
Costa-Neto, G., Galli, G., Carvalho, H. F., Crossa, J., & Fritsche-Neto, R. (2021). EnvRtype: a software to interplay enviromics and quantitative genomics in agriculture. G3 Genes|Genomes|Genetics, 11(4), jkab040. https://doi.org/10.1093/g3journal/jkab040
@article{costaneto2021envrtype,
title = {EnvRtype: a software to interplay enviromics and quantitative genomics in agriculture},
author = {Costa-Neto, Germano and Galli, Giovanni and Carvalho, Humberto Fanelli
and Crossa, Jos{\'e} and Fritsche-Neto, Roberto},
journal = {G3 Genes|Genomes|Genetics},
volume = {11},
number = {4},
pages = {jkab040},
year = {2021},
doi = {10.1093/g3journal/jkab040},
url = {https://academic.oup.com/g3journal/article/11/4/jkab040/6129777}
}From within R:
citation("EnvRtype")Please also kindly let the maintainer know when you are using EnvRtype (contact germano.cneto@gmail.com).
Released under the GPL-3 license — free and open source. You may use, study, modify and redistribute the code, provided derivative works remain under the same terms.