Evaluation of δ¹⁵N analysis to trace the origin of Diaphorina citri (Hemiptera: Liviidae) to citrus orchard fertilization management

Impact of developmental stage on Diaphorina citriδ¹⁵N values incorporation

Diaphorina citri adults 4 to 7 days old (age of D. citri adults referred as days after emergence) were obtained from a colony established at the University of California Riverside (UCR). This colony has been reared on M. koenigii plants since 2011 in environmental rooms (∼27 °C, ∼50% RH, 8:16 D:L, light source; Sylvania Octron 4100k fo32/741 32 w) in the Insectary and Quarantine Facility at UCR. Collected D. citri adults were sorted with a 1:1 sex ratio in a mating cage that contained M. koenigii seedlings on an environmental room with the same conditions specified above. These seedlings were sown and grown in approximately 1g mixture of Vermiculite and Pearlite (2:1) without fertilization treatment inside of an 18 × 150 mm borosilicate glass culture tube (Fisher Scientific, Hampton, NH, USA). Seedlings developed until their stem reached approximately 8 cm in height with more than 5 leaves. 100 D. citri adults 9 to 11 days-old were confined inside the cages for mating and oviposition for four days. After the confinement period, the adults were removed, and their offspring developed until adult stage on the seedlings. The resulting D. citri adults were used as the initial population (F₀) for the experiment. Seeds of M. koenigii were collected from two trees of the UCR Citrus Variety Collection. Upon collection, seeds were extracted from the fruit and washed to remove most of the extraneous flesh before being sterilized in a 2% NaClO solution. Thereafter, 300 seeds were split evenly into two groups in a greenhouse: the first group labeled as “organic” being fertilized with an organic OMRI certified bone meal fertilizer (Jobe’s Organics, 2-14-0 NPK), while the other group labeled as “conventional” was treated with Osmocote (14-14-14 NPK). The assigned fertilizer was applied (52 mg/pot) to the seeds upon sown and every three months until the M. koenigii plants were 6 to 8 months old.

The F₀ D. citri adults were divided equally between organic and conventional treatments and were exposed to six M. koenigii plants (initial plants) from each fertilization treatment for five days (in environmental rooms at ∼27 °C; ∼50% RH, 8:16 D:L, light source; Sylvania Octron 4100k fo32/741 32 w). After oviposition, the insects were removed and stored at −20 °C. The resultant offspring growing on each plant were removed from their initial plant/treatment at different stages of development (2nd instar nymphs, 4th instar nymphs and adults 0 to 1, 0 to 3 and 0 to 7 days old [∼20 D. citri nymphs or adults on each of the three plants assigned to a specific “D. citri stage/age transferred” namely as “group of psyllids” and treatment for a total of n = 30 plants]) and randomly transferred to a new M. koenigii plant with the opposite fertilization treatment (the individuals that hatch on the plants with organic treatment were transferred to plants with conventional treatment and the opposite) (Fig. S1). After being transferred to their new host plat/treatment, all D. citri individuals from the 30 M. koenigii plants developed to adulthood and continued to feed for 12 days after emergence on the last plant/treatment. On day 12, they were removed from the plant and stored at −20  °C. From each plant, two to six adults were selected to perform δ¹⁵N analyses (see ‘Plan material and D. citri sample δ ¹⁵N analysis). Three leaves per plant for a total of 42 plants (12 plants where the F₀ oviposited and 30 plants to which the insects were transferred) were processed to analyze the δ¹⁵N (see Plant material and D. citri sample δ ¹⁵ N analysis).

Relationship between the citrus leaf and D. citriδ¹⁵N values

Four citrus orchards (C1, C4, O1 and O4) (Table 1) were surveyed for the presence of D. citri populations, during the summer of 2017 (May-August), when D. citri populations usually peak in Southern California (Milosavljević et al., 2018). Ten to 20 trees in each grove with at least two D. citri colonies per tree were selected. White paint filter bags of 3.8 L (Workforce, China) were used as exclusion bags and secured with a zip-tie over each infested flush to isolate D. citri colonies with a maximum of three colonies per tree. Exclusion bags were checked weekly until more than 50% of the nymphs were either deceased or emerged as adults (71 exclusion bags with adults obtained), then the full branch was trimmed and stored at −20  °C until δ¹⁵N analyses of plant and insect material (see ‘Plant material and insect sample δ¹⁵N analysis’).

Field sampling for citrus leaf δ¹⁵N analysis

A total of 21 citrus orchards were sampled in the South and Central regions of California for δ¹⁵N analysis, consisting of 12 Valencia and 9 Navel sweet orange groves (Citrus sinensis (L.) Osbeck) all with Carrizo citrange as the rootstock (Table 1). Out of these 21 orchards, 11 were managed according to USDA organic guidelines and 10 were managed according to conventional methods (not organic) (Table 1). At each grove, a transect was selected to best encompass the presence of D. citri (checked by visual inspection) and the range in soil and environmental differences (a diagonal transect of the grove selecting a tree every other row when the D. citri presence allow it). Ten to 18 trees that fell upon this transect were randomly selected, and five young leaves (leaves from the most recent flushing period, light-dark green and fully expanded leaves) on different branches were collected at eye level (1.5 to 2 m above soil level) and stored at −20 °C until processed (see Plant material and D. citri sample δ¹⁵N analysis). Of these, we selected five trees randomly and three leaves for each were analyzed to compare leaves values of δ¹⁵N between conventional and organic orchards. Values of δ¹⁵N obtained from the plant material in the same citrus orchards but in different experiments from this study were also included in this analysis (see Relationship of the citrus leaf and D. citri δ¹⁵N values).

Six orchards (C1, C3, C4, O1, O3 and O4) were sampled on different seasons to assess seasonal variation of δ¹⁵N values of citrus leaves (Table 1). The six orchards were surveyed on March of 2017 (Spring). Additionally, the orchards C3, C4 and O3 were surveyed in December of 2016 (Winter) and C1, O1 and O4 were surveyed in July of 2017 (Summer).

Field experiments of this and the previous section were approved by the United States Department of Agriculture’s Animal and Plant Health Inspection Service (APHIS) (project number: #16-8130-0668-CA).

Plant material and D. citri sample δ¹⁵N analysis

Citrus and M. koenigii leaves were rinsed with deionized water three times before being dried at 60 °C in an Isotemp hybridization incubator (Fisher Scientific, Hampton, NH, USA) for 48 h. D. citri individuals were dried at 55  °C for 24 h or constant dry weight in the same incubator as the plant material. Leaves were individually ground into a fine powder using a mortar and pestle. D. citri samples were not ground and all material was stored separately in 2 mL microtubes. Target sample mass was calculated according to University of California, Davis Stable Isotope Facility (SIF) standards. Each sample was weighted in a 0.0001 g resolution scale (Intelligent Weighing Technology, Camarillo, CA, USA) then enclosed in 9 × 5 mm tin capsules (Elemental microanalysis, Okehampton, UK) and sealed to prevent leakage or contamination during the shipment. Samples were shipped to the University of California, Davis SIF in 96 well plates and analyzed using a PDZ Europa ANCA-GSL elemental analyzer interfaced to a PDZ Europa 20–20 isotope ratio mass spectrometer (Sercon Ltd., Cheshire, UK).

Statistics

All the statistical analyses described below were done in R (http://www.R-project.org). The difference between fertilization management on the δ¹⁵N values of the plant material in the experiment of δ¹⁵N values incorporation on the different D. citri stages were compared by one way ANOVA followed by Bonferroni’s posthoc tests for multiple comparison for all group of psyllids. The δ¹⁵N values for the insects in the same experiment were not independent since the source of all the insects for each treatment was from the same group of plants. Therefore, these values were analyzed using a linear mixed effects model with developmental stage (group of psyllids) as random factor with the “R” packages “car” and “nlme”. After this we ran pairwise comparisons between fertilizations for each group of psyllids using estimated marginal means with Tukey’s adjustment with the “R” package “emmmeans”. The average δ¹⁵N values of the citrus leaves and D. citri individuals from each exclusion cage installed on the citrus canopies were related with a linear regression. The citrus leaf δ¹⁵N values between seasons within the orchard were analyzed by Student’s t-test (see Citrus leaf δ ¹⁵N values on different seasons result). Finally, the δ¹⁵N values by leaf of the 21 citrus orchards sampled to assess differences between fertilization managements were analyzed by ANOVA, and the average of the citrus leaf δ¹⁵N values by orchard data were modeled using a kernel density estimation for the distribution of between-group (conventional and organic) variability with the “R” packages ggplot2, plyr and sm.

Impact of developmental stage on Diaphorina citriδ¹⁵N values incorporation

The average δ¹⁵N values of the 41 M. koenigii plants grown with organic (8.139 ± 0.199‰) and conventional (2.867 ± 0.197‰) fertilization treatments were significantly different (F_1,124 = 355.23, P < 0.001) (Fig. 1A, note that the initial M. koenigii plants, where the F₀ oviposit, for each treatment are included in the analysis but not in the figure). The plants from the organic treatment assigned to each group of psyllids did not differ significantly on its δ¹⁵N values (F_5,57 = 2.166, P = 0.071). However, the δ¹⁵N values from the plants with conventional treatment assigned to each group of psyllids differed significantly (F_5,57 = 6.989, P  < 0.001) (Fig. 1A).

The δ¹⁵N values from the F₀ D. citri adult population (those psyllids raised on non-fertilized plants and that laid the eggs on the fertilized initial M. koenigii plants) did not significantly differ after being fed, as adults, on conventionally or organically fertilized M. koenigii plants for five days (F_1,18 = 2.659, P = 0.12). However, their offspring, which spent some time feeding on fertilized plants as nymphs, had higher δ¹⁵N values (F₀ D. citri adults δ¹⁵N = 0.945 ± 0.307‰; n = 20).

The values of δ¹⁵N on the D. citri adults obtained were not influenced by the fertilization treatment (χ ² = 0.453, df = 1, P = 0.501) or the group of psyllids (χ² = 2.794, df = 4, P = 0.593), but the interactions between these two variables was significant (χ² = 123.384, df = 4, P < 0.001). The average δ¹⁵N values of D.citri adults raised on M. koenigii plants with conventional fertilization and transferred to plants with organic fertilization during the 2nd and 4th nymphal stages were higher (6.356 ± 0.551 and 7.154  ± 0.156‰  for 2nd and 4th respectively) compared to those same stage transferred from organically to conventionally managed plants (3.629 ±  0.247‰ and 4.752 ±  0.335‰ for 2nd and 4th respectively) (2nd: P < 0.001; 4th: P < 0.001) (Fig. 1B). However, D. citri adults that spent their entire nymphal period feeding on conventional or organic plants prior to being transferred to the opposite treatment 0–1 and 0–3 days after emerging as adults and feeding on their new hosts for 12 days, did not differ in the δ¹⁵N values between fertilization treatments (0–1 days: P = 1; 0–3 days: P = 0.844) (Fig. 1B). Conversely, D. citri adults transferred from organic plants to conventional plants at the stage “0–7 days old” had an average value of δ¹⁵N, 7.914 ± 0.274‰, higher than those transferred from conventional plants to organic plants at the same stage [5.098 ± 0.156‰, P  <  0.001] (Fig. 1B).

Relationship between the citrus leaf and D. citriδ¹⁵N values

The δ¹⁵N values for D. citri adults obtained in each exclusion cage were positively correlated with the δ¹⁵N values of the young citrus leaves from the same exclusion bag (Fig. 2). However, there were no significant influence of the fertilization managements (conventional vs. organic) on either the δ¹⁵N values of the leaves from the exclusion bags (avg. 6.774 ± 0.178‰, n = 271; F_1,67 = 0.623, P = 0.433) or on the δ¹⁵N values of the D. citri adult individuals reared inside those exclusion bags (avg. 7.272 ± 0.169‰  n = 258; F_1,67 = 1.379, P = 0.245). Therefore, it was not possible to trace the origin of the individuals. The difference between the average δ¹⁵N values of the leaves and the δ¹⁵N values of the D. citri adults per cage showed and an increase of 0.471 ± 0.125 (n = 71).

Citrus leaf δ¹⁵N values in different seasons

No significant differences were found in the citrus leaf δ¹⁵N values between seasons within orchards (C1; t(36) = 0.91, P = 0.437, C3; t(28) = 1.39, P = 0.176, C4; t(21) = 0.63, P = 0.534, O1; t(23) = 1.36, P = 0.186, O3; t(21) = 0.08, P = 0.936, O4; t(16) = 0.9, P = 0.381). The δ¹⁵N values of the 239 young citrus leaves analyzed from six orchards (C1, C3, C4, O1, O3 and O4) varied between a minimum of 2.352 ‰ in O1 in Summer of 2017 and a maximum of 18.393‰ in the same orchard in Spring of 2017. From the citrus orchards compared in winter (December 2016) and spring (March 2017) (C3, C4 and O3), O3 had the highest average citrus leaf δ¹⁵N value with 9.589 ± 0.261‰ and 9.557 ± 0.143‰ in winter and spring respectively (Fig. 3). From the citrus orchards compared in spring (March 2017) and summer (July 2017) seasons (C1, O1 and O4), the highest average citrus leaf δ¹⁵N values were obtained from O1 with 8.212 ± 1.035‰ and 9.029 ± 1.179‰ in Spring and Summer respectively (Fig. 3).

Citrus leaf δ¹⁵N values in conventional and organic groves

Significant differences were found in the δ¹⁵N values between fertilization managements (F_1,548 = 79.244, P < 0.001). On average, the organic and the conventional orchards had a δ¹⁵N value of 8.158  ± 1.038‰ and 5.185 ± 0.561‰ respectively. The maximum δ¹⁵N value (20.103‰) was detected in an organic orchard (O2) and the minimum δ¹⁵N value (1.215‰) was obtained from a conventional orchard (C8) (Fig. 4). Modeled Kernel distribution density curves for the citrus leaf δ¹⁵N values for the organic and conventional orchards showed that data from conventional orchards followed a normal distribution (Shapiro–Wilk normality test = 0.978, P = 0.956) and data from organic orchards followed a bimodal distribution (Fig. 5). Although there was a definite tendency for conventional citrus orchards to have lower δ¹⁵N values, there was overlap between the δ¹⁵N values for the organic and conventional citrus orchards (Fig. 5). Ninety-nine percent (p = 0.01) of the modeled conventional δ¹⁵N values would be expected to have a δ¹⁵N value below 10‰. This means that δ¹⁵N values above 10 ‰  could be expected to be obtained from organic citrus orchards, whereas δ¹⁵N values below 10 ‰ could not be unambiguously identified as organic or conventional.