The experiment was conducted during 2016 and 2017 at the West Florida Research and Education Center near Jay, FL (30°46'32.5"N 87°08'13.5"W, 62 m above sea level) on a Red Bay sandy loam (0-2% slope, fine-loamy, kaolinitic, thermic Rhodic Kandiudults) and one yr (2016) at the Plant Science Research and Education Unit near Citra, FL (29°24'28.9"N 82°08'43.4"W, 21 masl) on a Candler sand (0-5% slopes, hyperthermic, uncoated Lamellic Quartzipsamments) for three site-yrs.

Cultural practices were consistent with local production recommendations (Wright et al. 2016) with modifications as described below. All fields were fertilized and limed prior to planting according to soil test recommendations. Plots consisted of eight rows spaced 0.9 m apart and 7.6 m long. The experiments were planted on 27 and 10 May for 2016 and 2017, respectively. The experiments were strip-tilled into rye (Secale cereal L.) residue at Jay, FL and triticale (Triticosecale spp.) residue at Citra, FL. Seed of the peanut cultivar, Georgia-06G (Branch, 2007), was treated with azoxystrobin, fludioxonil, and mefenozam (Dynasty, Syngenta, North Carolina USA) as a fungicidal seed treatment. Final peanut stands were six plants per 0.31 m-row. Gypsum was applied at 2242 kg/ha 40 d after planting (DAP). Weeds were controlled as needed, avoiding the use of paraquat dichloride in order to obviate foliar damage and potentially affect foliar fertilizer uptake. Sites were irrigated as needed using overhead lateral irrigation.

A four (foliar fertilization) by two (digging date) factorial experiment was employed using a randomized complete block design with four replications. The two digging dates were approximately 2200 and 2500 aGDD, tracked using the aGDD model, representing early and optimal digging dates (Colvin et al., 2014; Rowland et al., 2006).

Foliar fertilization treatments were 10.0 kg N/ha as urea, 1.0 kg P₂O₅/ha as triple super phosphate (TSP), 0.34 kg B/ha (Max-In® Boron, Winfield Solutions, LLC, St. Paul, MN) during each application and a control. A low rate of N was applied in order to avoid leaf tissue damage (Nicoulaud and Bloom, 1996), though at the rate applied in this study, some leaf burn was observed at times. Applications were made twice during the season, one at R1 and again two wks later at R2 (Boote, 1982), corresponding to maximum nutrient uptake during early reproductive growth (Garcia and Hanway, 1976). Foliar fertilizer treatments were applied using an eight-row sprayer fitted with Cone-Jet TXVS-18 nozzles (Cone-Jet®, TeeJet Technologies, Spraying Systems Co., Wheaton, IL) at 187 L/ha (Bi and Scagel, 2007; Halevy et al., 1987). The first foliar application was made at R1 approximately 40 DAP (or 750 aGDD) and the second application at R2, two wks later (approximately 1000 aGDD). In 2016, foliar applications occurred on July 7 and 26 at Jay, FL. The one-wk delay in the second application during this site-year was necessitated by 85 mm precipitation between july 20^th to 22^nd 2016. Citra, FL applications were made on July 12^th and 25^th 2016. In 2017, applications were made on June 27^th and July 12^th in Jay, FL. Twenty-four hr after each foliar application, 25 leaf tissue samples were collected randomly in the middle two rows to quantify leaf tissue nutrient status. Leaf tissue samples consisted of leaflets and the petiole from second nodal leaves on the apex stem in a representative area in each plot. The leaves were washed, dried and ground using a Cyclone Lab Sample Mill (UDY Corporation, Fort Collins, Co) to pass a 1 mm sieve. Leaf tissue nutrient concentration was determined using a peroxide digestion followed by inductively coupled plasma mass spectrometry (Beauchemin, 2006).

During both years in Jay, normalized difference vegetation index (NDVI) was measured periodically throughout the experiment. NDVI is the amount of near-infrared and red light reflected by vegetation and is derived from the red near-infrared reflectance ratio (Pettorelli et al., 2005) and is often considered as being indicative of plant canopy photosynthetic area (Thelen et al., 2004). NDVI data were recorded weekly using a GeoScout GLS-400 (Holland Scientific Inc, Lincoln, NE) prior to first flower until after canopy closure. The sensor was placed approximately 60 cm over the canopy while walking through the experimental treatments. NDVI data were not recorded in Citra.

Digging dates were monitored using aGDD models (Anonymous 2013; Colvin et al., 2014; Rowland et al., 2006). Harvest treatments were targeted for 2200 and 2500 aGDD, representing early and optimal harvest timings, respectively (Rowland et al., 2006). The digging dates at Jay, FL in 2016 were 2184 aGDD (or 112 DAP, considered early) and 2504 aGDD (or 130 DAP, considered optimal), and in 2017, 2326 aGDD (or 117 DAP, considered early) and 2497 aGDD (or 132 DAP, considered optimal). The digging dates in Citra, FL during 2016 were 2430 aGDD (or 126 DAP, considered optimal) and 2650 aGDD (or 137 DAP, considered past optimal). The delay in digging dates at Citra were caused by 52 mm precipitation from Sept 12^th to 15^th 2016.

The four middle rows of the eight-row plots were dug using a four-row digger creating two windrows. One windrow was used to determine the percent mature pods and the second windrow was used to determine yield. To determine the percentage of mature pods, immediately after digging, all pods were removed from sampled plants until 180 to 220 pods were collected. Pods were blasted using a pressure washer with a turbo nozzle to remove the exocarp and expose the mesocarp color (Carter et al., 2017). The maturity profile was determined using two methodologies described below. One classified pods as mature and immature based on the mesocarp color using a peanut profile board (Carter et al., 2017; Williams and Drexler, 1981). Brown and black pods were considered mature. White, yellow, and orange pods were not used for analysis to ensure only the most mature pods were considered in maturity evaluations. The second method scanned blasted pods on a digital scanner (HP OfficeJet 7612, Palo Alto, CA). The resulting images were analyzed using the PeanutFARM (Anonymous 2013) tool to quantify the percentage of brown/black (mature) pods (Colvin et al., 2014). Peanut pod yield was recorded using the second windrow, approximately one wk after digging, using a plot combine. Final yield data were adjusted to 10% moisture.

Repeated measures analysis of variance for a randomized complete block was conducted for leaf tissue nutrient, NDVI, and yield data using the PROC MIXED procedure in SAS (Version 9.4, SAS Institute Inc., 2017) at 95% confidence level unless otherwise indicated. Individual analyses were performed for each site-year. For leaf tissue nutrient analyses, foliar fertilizer and application time were considered fixed effects. For yield analyses, foliar fertilizer and harvest time were considered fixed effects. NDVI analyses considered foliar fertilizer and time of observation as fixed effects. In all cases, replication and its interactions were considered random effects. Pairwise least square means were separated using least significant differences (LSD) at the 95% confidence level with the %PDMIX800 macro (Saxton, 1998) within SAS 9.4. Figures were generated using R software version 3.6.0 (R Core Team, 2019).

NDVI values did not differ among foliar fertilizer treatments (data not shown), indicating very little effect of added nutrients on canopy condition. However, NDVI is most effective at detecting nutrient impacts when deficiencies are present. For example, nutrient deficiencies have been detected by NDVI measurement of crops such as soybean and corn (Zea mays L.) (Milton, et al., 1991; Osborne, et al., 2002; Thenkabail, et al., 2000). The lack of differences in NDVI values in the present study indicate that nutrient deficiencies were likely not present so that foliar fertilization had little relative impact on plant nutrient status (Bryson and Mills, 2014). NDVI did differ by aGDD as the crop developed, a common result in other studies since NDVI increases with leaf area index as the crop matures (Elvidge and Chen, 1995; Huete et al., 1985).

Even though foliar fertilization did not affect NDVI, direct measures of leaf tissue N concentrations were affected by fertilization treatments (Table 1). Compared to the control, leaf tissue N measured 24 hr after foliar N application increased during two of three site-years. The lack of consistent response in leaf tissue N concentration may be due to the relatively quick translocation of N from leaves to fruits (Wittwer et al., 1963). Uptake and translocation toward the developing seed may have already occurred when the leaves were sampled 24 hr after application, particularly given the relatively low amount of foliar N applied.

After foliar P application, leaf P concentration increased during two of three site-years compared to the control (Table 1). These results are somewhat more consistent than those reported by Walker et al. (1982), who generally noted no leaf tissue P response after foliar P application, even at rates up to 5.8 kg P/ha. However, that study sampled tissue 10 to 14 d after application and every 28 d thereafter, whereas the current study sampled tissue 24 hr after application. The shorter interval between application and sampling may explain the greater tissue P response compared to the control in the present study.

Boron fertilization consistently resulted in an increase in leaf tissue B concentration compared to the control (Table 1) in Jay and Citra after each application. The high tissue B concentrations indicated the nutrient was absorbed by the leaf. Konsaeng et al. (2010) reported that upper foliar B concentration decreased with time concomitant with an increase in lower foliage B concentration, demonstrating a remobilization of B during a 13-d span.

In 2016 at Citra, FL, yield was not affected by foliar treatment (Figure 1). This location was harvested at optimum maturity (2430 aGDD, or 126 DAP) and 150 aGDD past-optimum (at 2650 aGDD, or 137 DAP), though it bears noting that 137 DAP would be considered within the optimal harvest range by most growers. Furthermore, Georgia-06G is a medium maturity genotype, considered ready for harvest at 135 to 140 DAP (Branch, 2007; Branch and Brenneman, 2009). The 2016 Jay, FL site was harvested at both an early (2184 aGDD, or 112 DAP) and optimum digging date (2504 aGDD, or 130 DAP), resulting in lower yield during the earliest harvest date (p=0.0004, Figure 1). Yield increased by 560 kg/ha during the early digging date in Jay when foliar P was applied compared to the control, but this effect was significant only during one of the three site-yrs. Foliar applications of B or N did not increase yield at any digging date, nor did P application when peanut was dug during an optimal time. In 2017, yield was lower (Figure 1) at the early harvest date (2326 aGDD, or 117 DAP) compared to the later harvest (2497 aGDD, or 132 DAP). During both harvest timings, foliar treatments did not affect yield.

Within harvest date, the percentage of mature pods was not generally affected by foliar fertilization treatments during any site-year using both methodologies (Figure 2), except for boron foliar fertilizer which increased the percentage of mature kernels assessed by profile board during the second digging date in Jay 2016. As expected, the percentage of mature pods increased when pods were dug at optimal timing compared to early digging, except in 2017 using the digital imaging method. Maturity assessment using the profile board showed more variation than the digital imaging method, likely because immature pods can often be smaller than mature pods. Results from this study did not support the hypothesis that foliar fertilization affects the peanut maturity profile under these conditions. However, foliar fertilizers might influence the maturity profile of peanut, under nutrient limiting conditions.