Fig 1.
Workflow of the modeling process used in this study.
Occurrences were collected, cleaned, and employed to estimate three model calibration regions (i.e., Mi, Mg, and Md). Present-day climatic variables were restricted to these model calibration regions and compared to future climatic conditions in Minnesota. Models were parametrized using present-day climates in the three model calibration regions and the best models were projected to future climates in Minnesota using five climate models and four RCP scenarios.
Fig 2.
Model calibration region, M, explored in this study.
Models were calibrated in three regions (red lines in A, B, and C) based on the distribution of starry stonewort populations (green points). A. Model calibration region based on an invasive population approach focused on starry stonewort populations in the invaded area of the United States and a high dispersal potential (i.e., 2,200 km), Mi. B. Model calibration region considering the entire or global species’ range in the United States, Europe, and Japan and a high dispersal potential (i.e., 2,200 km), Mg. C. Model calibration region considering the entire or global species’ range in the United States (left map), Europe (central map), and Japan (right map) and a reduced dispersal potential (i.e., 700 km), Md.
Fig 3.
Ecological niche model transference to Minnesota under present-day climate.
Ecological niche model predictions based on model calibration region in the invaded range with high dispersal (Mi; top), entire species’ range with high dispersal (Mg; mid), and entire species’ range with reduced dispersal (Md; bottom) projected to Minnesota to identify areas with high (red) or low (blue) environmental suitability (left) and high (pink) or low (light blue) model uncertainty (right).
Fig 4.
Environmental similarity comparison between the calibration Mi and the projection region of Minnesota.
Exdet tool identified analogous climates between present-day climate in the calibration region from the invaded range and future climate scenarios in the projection region of Minnesota. Areas with analogous (green) and non-analogous environments in Minnesota (grey) were identified for five future climate models (i.e., CCSM, GISS, IPSL, MIROC, MRI) and four RCP scenarios of CO2 emissions (i.e., 2.6, 4.5, 6, and 8.5).
Fig 5.
Environmental similarity comparison between the calibration Mg and the projection region of Minnesota.
Legend as in Fig 3.
Fig 6.
Environmental similarity comparison between the calibration Md and the projection region of Minnesota.
Legend as in Fig 3.
Fig 7.
Ecological niche models of starry stonewort calibrated in Mi and projected to future climate scenarios in Minnesota.
Ecological niche model predictions based on model calibration region Mi projected to Minnesota. Areas with high (red) or low (blue) environmental suitability (Suitability, left) and high (pink) or low (light blue) model uncertainty (Standard deviation, right) were identified for five future climate models (i.e., CCSM, GISS, IPSL, MIROC, MRI) and four RCP scenarios of CO2 emissions (i.e., 2.6, 4.5, 6, and 8.5).
Fig 8.
Ecological niche models of starry stonewort calibrated in Md and projected to future climate scenarios in Minnesota.
Ecological niche model predictions based on model calibration region Md projected to Minnesota. Legend as in Fig 7.
Fig 9.
Ecological niche models of starry stonewort calibrated in Mg and projected to future climate scenarios in Minnesota.
Ecological niche model predictions based on model calibration region Mg projected to Minnesota. Legend as in Fig 7.
Fig 10.
Starry stonewort future climate models ensemble.
Model ensemble expressed as the average of continuous models in logistic format (left, ‘Suitability’), showing areas with high (red) or low (blue) suitability from all the RCP emission scenarios in comparison with the maximum range of suitability of climatic models projected to Minnesota in present environmental conditions (i.e., from the lowest [0.09] to the highest [0.69] suitability). Lack of agreement was estimated from the standard deviation of the final models (right, ‘Standard deviation’) and shows areas of high (pink) or low (light blue) disagreement among models. Top: Models calibrated in Mi and projected to future climate scenarios in Minnesota. Mid: Models calibrated in Mg and projected to future climate scenarios in Minnesota. Bottom: Models calibrated in Md and projected to future climate scenarios in Minnesota.
Fig 11.
Conceptual framework used for interpretation of predictions.
Top: The “Hutchinson Fallacy” expressed as the intersect of abiotic (A; dashed line) and biotic factors (B; dotted line) showing the environments that a species can occupy (gray) or not (white area inside the dotted circle), based on biotic interactions solely (e.g., competitors). Note that under the Hutchinson’s proposal, all the areas environmentally suitable can be reached by the species (i.e., entire circle), suggesting that the movement and dispersal potential of the species (M; solid line) is effective to occupy all the suitable conditions (i.e., A is contained in M). Bottom: The “BAM Framework” proposed by Soberón and Peterson [23] to explain that dispersal limitations (M) can also restrict the species to occupy (gray) only a portion of all the suitable environments (A). Note that in this example, the species can occupy a portion of the environmental conditions suitable due to the limited dispersal potential (i.e., half circle). A (abiotic) = environmental conditions suitable for the species; B (biotic) = interaction with other species; M (move) = movement or dispersal potential of the species.