Predicting Soil Phosphorus Requirements

Author : Stewart Gordon Wilson
Publisher :
ISBN 13 : 9781658416931
Total Pages : pages
Book Rating : 4.4/5 (169 download)

DOWNLOAD NOW!

Download or read book Phosphorus in California written by Stewart Gordon Wilson and published by . This book was released on 2019 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: This work focuses on the distribution, management and pedological underpinnings of phosphorus biogeochemistry in California. Phosphorus (P) is a critical element to the function of life on earth, an essential component of a myriad of biomolecules, a vital plant nutrient, and a non-renewable resource. Its distribution defines plant communities, its deficiency drives symbiosis, and its availability is critical for global food security. The multiple negative charges of phosphate, that drive its utility as a critical biochemical element, also make it a dynamic biogeochemical element, with P binding strongly to soil colloids such as Fe/Al-(hydr)oxides and poorly crystalline allophane and imogolite. This work leverages the soil forming factors to predictively map soil P in California vineyard soils, investigates the influence of climate and lithology on soil P fractions, and highlights compost application strategies in a truncated California vineyard soil thought to be P fixing. Chapter one investigates composts, and compost application rates, to improve vine performance in a degraded vineyard soil in the Red Hills American Viticulture Area (AVA) of North Coast California. Previously, chemical fertilization had not remediated the underperforming vines. We sought to use a soil health, or soil system approach to improve vine performance, as composts can be expected to have biological, physical and chemical benefits to soils. This chapter highlights a two-year experiment on the effect of compost application rate on soil health, vine growth, harvest characteristics (berry weights, per vine yields and cluster weights), and juice quality in a cabernet sauvignon vineyard. Compost was applied at three rates (11.2, 22.4 and 33.6 t/ha) as a single dose before the 2012 growing season. Petioles were sampled at bloom, per vine yields, berry weights, cluster weights taken at harvest, and pruning weights and soils sampled post-harvest in the 2012 and 2013 growing seasons. Berry weights, pruning weights and yields increased with compost application rate. Compost improved most soil properties, and reduced P sorption. Vine petiole nutrients (N,P,K) were significantly increased from compost application in both years. Vine balance (ratio of pruning weights to per vine yield) and juice characteristics (pH, total soluble solids and titratable acidity) were not affected by compost application rate. We conclude that significant multi-year benefits to soil health, vine nutrient status, yields and vine performance can be achieved in degraded vineyard soils from a single dose of compost at a higher application rate (22.4 and 33.6 t/ha) without compromising juice characteristics or vine balance. Chapter two investigates how soil forming factors, particularly climate and lithology, influence P biogeochemistry. The dominant paradigm of the fate of P during pedogenesis has focused on one pedogenic state factor, time, as the driving force of P distribution. The objective of this study was to integrate climate, and especially lithology, into the concept of P transformations across a pedogenic gradient. We suggest that the fate of P is not only linear with time, but multidimensional integrating all pedogenic state factors. We investigate P fractions and pedogenic Fe and Al across three bedrock lithologies spanning 2150 m in elevation and four bioclimatic zones corresponding to the dominant tree species, red fir (RF), white fir (WF), ponderosa pine (PP), and oak-woodland (OK). Soils become more weathered descending the gradient, with peak Fe-(hydr)oxide production in mid-elevation sites (PP). In basalt and andesite, colloids transition from poorly crystalline Fe-(hydr)oxides and allophane/imogolite to more crystalline Fe-(hydr)oxides and kaolin clays descending the transect. Parent materials represent a large gradient of P contents, ranging from 131 mg P per kg in granite to more than 1500 mg P per kg in andesite. This range of parent material elemental compositions, as well as the range of degrees of chemical weathering and suites of clay minerals, allowed for a unique investigation of the effect of climate and lithology on the fate of soil P across a pedogenic gradient. Lithology significantly influenced most P fractions. For example, granite soils were uniformly high in labile inorganic P (Pi), despite having the lowest rock P, owing to much lower Fe-oxide production. Ca-Pi (presumable apatite) declined in basalt and andesite with increased weathering, but was unchanged in granite, leading to granite being higher in Ca-Pi than all other rock types in the zone of intense weathering (PP). P occluded in metal oxides or recalcitrant organic forms provide the best example of differential effects of the weathering gradient on P dynamics. P occlusion increased with increased weathering in basalt, declined with increased weathering in andesite and was unchanged in granite. Declines in P occlusion with increased weathering in andesite is contrary to the dominant paradigm of progressive P occlusion with increased weathering and is due to conversion of high surface area poorly crystalline colloids to more crystalline Fe/Al-(hydr)oxides and kaolin minerals with increased weathering. In basalt, advanced chemical weathering leads to extensive crystalline Fe-(hydr)oxide production and greater P occlusion. Granite was much less chemically weathered, had relatively less pedogenic Fe-(hydr)oxide production and lacked appreciable poorly crystalline colloids, which led to lower P occlusion. Chapter two findings highlight the differential effect of climate on disparate rock types, leading to differential chemical weathering effects, different suites of clay minerals, and distinct differences in the effect of pedogenesis on P fractions. Chapter three takes a predictive soil mapping (PSM) approach to map P availability and sorption in California vineyard soils. This chapter uses soil samples, raster-based proxies for soil forming factors, soil attributes from the USDA soil survey database (SSURGO), three machine learning algorithms (MLA), and two different MLA ensemble techniques to investigate the distribution of P availability and sorption in California vineyard soils. 141 pedons were sampled across California vineyards and analyzed for Olsen P and a P-sorption index (PSI). Sampled soil attributes were aggregated by two depth weighted averages (0-30 cm and 30-100 cm). Base models included random forest (RF), extreme gradient boosting (XGB) and Cubist. Models were ensembled with two methods, an arithmetic combination of base models weighted by their R2 or a linear combination of base models via a generalized linear model. Ensembled models generally outperformed base models. Both depths (0-30 cm and 30-100 cm) of PSI were effectively modeled with a PSM framework (R2=0.68-0.73). Both depths of Olsen P were less effectively modeled (R2=0.46-0.56). Mapped predictions highlight regional trends in P distribution and potential differences in regional nutrient management strategies. Areas of high P sorption and acute P deficiency were predicted in upland volcanic soils of the Sierra Foothills AVA and North Coast AVA. In areas of Lodi AVA, soils were P deficient but only moderately P fixing, whereas in the volcanic uplands of Napa AVA soils were both P deficient and P fixing. This highlights site specific nutrient management strategies, with lower doses of P fertilizer required to overcome P deficiency in moderately P fixing Lodi soils versus highly P fixing soils of the Napa AVA. Model predictions of Olsen P revealed significant areas of critically deficient subsoils, highlighting the importance of soil conservation in these regions. P sorption was very effectively modeled with the PSM framework. P sorption is tied to Fe/Al-(hydr)oxides, which are strongly driven by the interaction of soil forming factors. Soil properties that can be tied to soil forming factors, such as PSI, can be more effectively predicted with a PSM framework than more dynamic soil properties, such as Olsen P, that are strongly driven by management. Results of this chapter highlight differences in regional nutrient management strategies and further contemporary machine learning based approaches to predictively map soil P.