Effects Of Environmental Factors On Plant Invasion And Risk Assessment For Invasive Species

Exotic species threaten regional biodiversity because they modify ecologicalcommunity structure, alter natural habitats and affect ecosystem function. Invasionecology focus on both the invasibility of habitats and the invasiveness of species:Which habitats are most susceptible to invasion? Which species are most likely tobecome invasive? Despite of extensive research, it is still difficult to identifyparticular traits and factors that are consistently associated with plant invasivenessand habitat invasibililty. Therefore, it is necessary to explore factors affectingcommunity resistance to exotic invasion in particular ecosystems and predict theirpotential variation in context of future climate change. On the other hand, it is alsonecessary to develope weed risk assessment models based on machine learningalgorithm in particular regions to predict potential invasion.In this thesis, we selected a mangrove ecosystem as a model endangered habitatto investigate the possibility of reducing the potential for invasion by altering lightavailability; we also selected North American coastal prairies ecosystem as the othermodel habitat to explore the potential effects of global warming and hurricanes withsalt inundation on tallow invasion into coastal prairies. In addition, we developed four risk assessement models with machine learning algorithms for invasive herbaceousplants in China, and screened out some important traits associated with the tendencyof plant speices to invade. Knowing this would in principle allow land managers tosingle out invasive species for early eradication in China.1. Compared with native Aegiceras corniculatum, S. apetala showed asignificantly higher growth rate for both height and biomass accumulation under fullirradiation. Compared to the full irradiation treatment, the shading treatmentsignificantly reduced the height, total biomass and biomass allocation to leaves of S.apetala by61.31,71.0, and76.2%, respectively, whereas the growth of A.corniculatum was not affected. The results suggested that lowering light availabilitycould inhibit the growth of S. apetala and increase the competitiveness of A.corniculatum. Therefore, the reduction of light avalibility under canopy can decreasethe invisibility of mangrove ecosystem and further increase its resistance to potentialexotic invasion. By taking advantage of the differences in shade tolerance betweenfast-growing exotic mangroves and native mangroves, introduction fast-growingmangroves in coastal area could be an effective strategy for preventing potentialinvasion and restoring wetland habitats.2. Soil warming had little effect on native coastal prairies and exotic tallow treeinvasion, whereas hurricanes with salt inundation showed great impacts. Saline watersignificantly reduced the height and ratio of root to shoot (R/S) of tallow tree.Biomass acclimation of tallow tree was only inhibited when soil salinity increasedwith soil warming. On the other hand, saline water dramatically inhibited the seedlinggrowth of little bluestem, and reduced their height, dominance, above-, andbelow-ground biomass. Therefore, higher soil salinity had a greater impact on nativeplant growth than they did on tallow trees. Future frequent hurricane may acceleratetallow tree invasion by increasing the invisibility of coastal prairies. In addition,hurricanes with salt inundation also significantly changed the community structure ofprairies community. Higher soil salinity increased dominance of C3-forb, whereas decreased the dominance of C4-grass little bluestems. The dominant species in prairiestransformed from little bluestem to Ragweed and Spartina spp. These results indicatedthat global warming and hurricanes with salt inundation might result in an ecosystemtransformation from coastal prairies to coastal swamp. Common Ragweed, a typicalagricultural weed, will overwhelmingly spread.3. We developed weed risk assessment models for invasive herb seed plants inChina through four machine learning algorithms: Classification and Regression Tree(CART), Multivariate Adaptive Regression Spline (MARS), Random Forest (RF) andMultiple Additive Regression Tree (MART). Distribution area in native range, seedgermination rate and seed mass were three most important traits screened out by thesemodels, showing strong correlation to herbaceous plant invasiveness. Exotic herb seedplants introduced to China are most likely to become invasive if they are widelydistributed in their native range or have higher seed germination rate and small seedmass. This demonstrated that climatic affinities of species and seed traits are usefulfor quantifying the risk of herbaceous plant becoming invasive. Risk assessments withgood performance were generated for all four methods (AUC ranges from0.79to0.88), which had overall accuracy from67.51%to80.17%. Random forests (RF) andMultiple Additive Regression Tree (MART) consistently produced tools with higherperformance and could be used as a standard approach to risk assessmentdevelopment.