Preliminary observations.

Extensive C. micronesica seed toxicology studies were ongoing when the A. yasumatsui invasion occurred. The armored scale immigrated into our research site in western Guam in Feb. 2005, and our ability to continue meeting the objectives for these studies was destroyed by the invasion. At this time, we had stored preinvasion seed tissue from several ontogeny and storage studies (Marler and Shaw, 2009, 2010; Marler et al., 2006). These samples were frozen at –40 °C and lyophilized in Dec. 2003 and Jan. 2004 before storage and were used to provide a look at the preinvasion seed carbohydrate status.

We were also monitoring a population of female trees with megastrobili that were 3 months in age when the A. yasumatsui infestations were initiated. The trees which provided the preinvasion stored seed tissue, and these postinvasion infested trees were growing in the same habitat with coralline soils formed in slope alluvium, loess, and residuum overlying limestone (Clayey-skeletal, gibbsitic, nonacid, isohyperthermic Lithic Ustorthents) (Young, 1988). We continued to monitor these scale-infested seeds until they began abscising at 19 months in age. Seeds were harvested from these trees at this time to provide a look at the postinvasion seed carbohydrate status.

Ten seeds were harvested from each of 10 postinvasion trees on 18 June 2007. Each seed was separated into outer soft integument tissue (sarcotesta), inner hard integument tissue (sclerotesta), and gametophyte tissue. The samples from each of the three seed tissues were combined for each tree into a composite sample to provide 10 replications of each tissue. The tissue was frozen at –40 °C then lyophilized.

We used stored sarcotesta, sclerotesta, and gametophyte tissue samples from seeds that were 19 months in age to ensure seeds of the same age were compared for pre- and postinvasion trees. Tissue from 10 stored samples that met these requirements was included. Methods for carbohydrate quantification followed Marler and Dongol (2016a). All tissue samples were milled to pass through 20-mesh screen. Soluble sugar extraction was conducted using hot-water extraction with acetonitrile (80 °C; Schloter et al., 2005). The concentrations of four free sugars (the hexoses fructose and glucose, and the disaccharides sucrose and maltose) were determined by HPLC-RI (Thermo Scientific RI-150, AS3000 autosampler, P2000 pump). Starch was quantified using the anthrone colorimetric method following removal of sugars by ethanol then hydrolysis of starch to glucose using sulfuric acid at 95 °C (Clegg, 1956).

No statistical test was applied to these observational data because there were too many confounding variables. The means and standard errors were compared for preinvasion (13 months before initial scale infestation) and postinvasion (16 months after initial scale infestation) samples to provide a preliminary determination of seed carbohydrate responses to A. yasumatsui.

Whole-tree protection.

Several plots throughout Guam were established in 2007 for protection from A. yasumatsui threat by using root drenches of imidacloprid every 3 to 4 months (Marler, 2018). A site on the east coast of Guam containing the same soil series described in the preceding text was used to determine seed carbohydrate status with and without plant protection. Synchronized reproductive events that emerged in Nov. 2011 were used to select the experimental female trees. Twelve healthy trees within the protected plot were selected, and 12 nearby unprotected trees with a synchronized megastrobilus were also selected. The trees ranged in height from 2.0 to 2.4 m. Seed set was greater for the healthy trees, so developing seeds were thinned on 24 Feb. 2012 to 15 seeds per megastrobilus among all of the replications to ensure homogeneous seed load throughout seed maturation. Determination of pollinated seeds based on increased size is an unambiguous endeavor at this age (Marler and Dongol, 2011).

Seed harvests occurred in June 2013 when seeds were 19 months in age. Four of the unprotected trees died during the study, so the protected tree closest to each of these trees was also removed from the study. Ten of the healthiest seeds were harvested from each of the eight remaining trees per treatment. The healthy seeds from the protected trees had received whole-tree insecticide protection for 6 years, and the infested seeds from outside the plot were harvested from trees that had been damaged by A. yasumatsui for 7 years.

The separation of seeds into three tissue categories was conducted as previously described. The fresh tissue was immediately placed in a forced draft oven at 75 °C and dried to a constant weight. The dry weight of gametophyte, sarcotesta, and sclerotesta tissue was determined for each 10-seed composite sample. Carbohydrate quantification was as previously described. The total pool of each carbohydrate was calculated for each tissue category by multiplying the concentrations by the corresponding tissue weight.

The two sources of variation were tree protection (two treatments) and seed tissue category (three categories). An exploratory analysis of the concentration and total pool data indicated many differences in variances due to treatment and tissue. Therefore, the Proc MIX function in SAS (SAS Institute, Cary, NC) was used for analysis, and the model allowed for differences in treatment variances and differences in tissue category variances while assuming that the measurements on the three tissues were correlated. Treatment design was a two-way factorial in a split plot design with treatment as the whole plot treatment and tissue as the subplot treatment. Means separation for significantly different means was conducted with Tukey’s honestly significant difference test.

Seed-only protection.

Twenty unprotected C. micronesica female trees with synchronized emergent reproductive structures were selected from the west coast experimental site in Nov. 2011. The naked ovules were protected every 2 weeks with horticultural oil spray to keep the A. yasumatsui infestation to a minimum but to allow natural pollination. Determination of seed set was unambiguous by 25 Feb. 2012, and the number of developing seeds per sporophyll was standardized to two by thinning individual seeds and unpollinated ovules. Moreover, when a tree exhibited an odd number of sporophylls, one sporophyll was removed to enable an even number of sporophylls for each of the two treatments.

Each sporophyll was covered by an individual sleeve constructed from lucite screen with 130 holes per cm and affixed to the base of the sporophyll. The sleeve for half of the sporophylls on each tree was left open to enable direct A. yasumatsui infestations on the seeds. The sleeves for the other half of the sporophylls on each tree were fully closed to exclude incoming A. yasumatsui crawlers. The few scale individuals that were observed inside of the closed sleeves were treated with Pyriproxyfen. This insect growth regulator is absorbed by the plant tissue but it is not systemic. A follow-up spray of horticultural oil suspension was applied to the closed sleeves after 2, 4, and 6 weeks to ensure mortality of the few initially enclosed A. yasumatsui individuals. Inspections for the duration of the study confirmed there were no A. yasumatsui individuals infesting the maturing seeds within the enclosed sleeves.

Six of these unhealthy trees died during the study. Moreover, most of the seeds on six of the 14 live trees were aborted due to compromised plant health. Protected and unprotected seeds were among the aborted seeds. These trees were excluded from the study. The remaining eight trees ranged in height from 2.2 to 2.6 m, and all seeds were harvested and separated into protected vs. unprotected seeds for each tree.

The separation of seed tissues and tissue preparation and chemical analyses were as described for the whole-tree protection study. These data met prerequisites for parametric approaches. The concentration and total seed pool data were subjected to a 2 × 3 factorial with two levels of seed protection and three tissue categories using the Proc GLM function in SAS. Means separation for significant variables was as described earlier.

Preliminary observations.

Cycas micronesica seed tissue carbohydrate concentrations either declined after the A. yasumatsui invasion or remained unchanged among the tissue categories (Fig. 2). The pre-invasion gametophyte concentrations of glucose, maltose, and sucrose were about double those of the postinvasion gametophyte tissue. Postinvasion gametophyte starch concentration was approximately two-thirds that of preinvasion gametophyte tissue. The gametophyte fructose concentration was not influenced by the scale invasion. All five carbohydrates declined following the invasion within sarcotesta tissue, with fructose concentration declining relatively more than the other carbohydrates. Sclerotesta tissue revealed the least amount of change in carbohydrate concentrations. The disaccharides maltose and sucrose declined after the invasion, but the other three carbohydrates were not influenced by the invasion.

View Full Size

The concentration of nonstructural carbohydrates in gametophyte (Gam), sarcotesta (Sar), and sclerotesta (Scl) tissue of 19-month-old seeds from Cycas micronesica trees on Guam 13 months before the invasion of A. yasumatsui (shaded bars) and 16 months after initial A. yasumatsui infestation (open bars). (A) Fructose, (B) Glucose, (C) Maltose, (D) Sucrose, (E) Starch. Mean ± SE, n = 10.

Citation: HortScience horts 54, 10; 10.21273/HORTSCI14235-19

• Download Figure
• Download figure as PowerPoint slide

The sum of all five seed carbohydrates after the A. yasumatsui invasion was 60% of the sum before the invasion for sarcotesta, 66% of the sum for gametophyte, and 79% of the sum for sclerotesta. These observational results confirmed that nonstructural carbohydrates declined relatively more in sarcotesta tissue than in the other tissues and illuminated that a manipulative experiment was needed to understand more fully the seed carbohydrate responses to direct herbivory by A. yasumatsui.

Whole-tree protection.

A significant interaction in fructose concentration occurred because the scale invasion reduced fructose to a greater degree in sarcotesta than in the other tissues (Table 1). Fructose concentration was greatest in the sarcotesta tissue and was similar for gametophyte and sclerotesta tissues. The main effects of invasion and tissue were both significant for glucose concentration, but the interaction was not significant due to a similar reduction in glucose among the tissues following the invasion. The invasion reduced the overall glucose concentration from 14.5 to 10.6 mg·g⁻¹ (mean of the three tissues). A significant interaction in maltose concentration occurred because the scale invasion did not influence concentrations in gametophyte or sclerotesta tissue but drastically reduced concentration in sarcotesta tissue. Maltose concentration of sarcotesta tissue was almost triple that of the other two tissues before the invasion, but maltose concentration was similar among the tissues after the invasion. A significant interaction in sucrose concentration occurred because the invasion reduced sucrose in gametophyte and sarcotesta tissues but did not influence sucrose in sclerotesta tissue. Sucrose concentration was greatest in sarcotesta and least in gametophyte tissue.

Table 1.

The concentrations and total pool size of nonstructural carbohydrates in gametophyte, sarcotesta, and sclerotesta tissue of mature seeds of healthy Cycas micronesica trees protected from Aulacaspis yasumatsui infestations by 6 years of imidacloprid insecticide applications and unprotected trees following 7 years of A. yasumatsui infestations. n = 8.

View Table

Starch concentration greatly exceeded that of the free sugars, and was greatest in gametophyte and similar in the other two tissue categories (Table 1). A significant interaction occurred for starch concentration because the invasion greatly reduced starch in gametophyte, slightly reduced starch in sarcotesta, and did not significantly influence starch in sclerotesta.

The sum of all five seed carbohydrates after the A. yasumatsui invasion was 63% of the sum before the invasion for sarcotesta, 78% of the sum for gametophyte, and 108% of the sum for sclerotesta. These results corroborated the preliminary observational data to show that nonstructural carbohydrate concentration declined relatively more in sarcotesta tissue than in the other tissues.

A significant interaction occurred for total pool of fructose, glucose, maltose, and sucrose, and these sugars exhibited similar relative patterns among the tissues. The preinvasion pool size and the magnitude of postinvasion reduction in pool size followed the order sclerotesta < gametophyte < sarcotesta for each of the sugars (Table 1). The substantial amplitude of preinvasion differences of these sugars among the tissue categories was muted by the invasion, as sugar pool size converged. The pool size of sucrose exceeded that of the other three sugars in every invasion × tissue category.

A significant interaction occurred for total starch pool because the invasion caused a drastic reduction in starch of gametophyte tissue, a moderate reduction in starch of sarcotesta tissue, and no change in starch of sclerotesta tissue (Table 1). The majority of seed starch was located in the gametophyte tissue before and after the invasion.

The pre- and postinvasion differences in carbohydrate pools were much greater than the differences in carbohydrate concentration. Therefore, the influence of direct infestation of seeds by A. yasumatsui was mostly a consequence of a reduction in seed biomass accumulation, and secondarily a consequence of reduced concentrations.

Seed only protection.

The sugar and starch concentrations of clean vs. infested seeds from unhealthy infested female C. micronesica trees (Table 2) were generally below those of seeds from the whole-tree protection study (Table 1). However, the patterns of significance and the rankings of the means were remarkably similar for the two manipulative studies, despite the contrast in whole-tree health. The primary difference in patterns was more overlap among the groups of means for starch concentration data but not for total pool data.

Table 2.

The concentrations and total pool size of nonstructural carbohydrates in gametophyte, sarcotesta, and sclerotesta tissue of 19-month-old seeds on unhealthy Cycas micronesica trees infested with Aulacaspis yasumatsui. Clean seeds were protected by lucite screen and remained pest-free; infested seeds were exposed to direct A. yasumatsui infestations. n = 8.

View Table

The sum of all five seed carbohydrates after the A. yasumatsui invasion was 65% of the sum before the invasion for sarcotesta, 72% of the sum for gametophyte, and 103% of the sum for sclerotesta. These results again corroborated the previous studies to show that non-structural carbohydrates declined relatively more in sarcotesta tissue than in the other tissues.

Future directions.

Our revelations that reduced seed resources may partly explain poor germination and seedling performance of infested seeds do not exclude other causal factors. For example, epigenetic regulation of gene expression is heavily involved in seed-to-seedling transition (Xu et al., 2018), and direct herbivory of seed integument tissues by the nonnative A. yasumatsui may directly alter these epigenetic phenomena. The study of the epigenetic basis of cycad seed development may allow for a better understanding of how A. yasumatsui herbivory negatively influences Cycas plants.

Our prediction that the exposed sarcotesta tissue would decline to a greater degree in carbohydrate resources during A. yasumatsui herbivory than the other seed tissues was confirmed for relative percentage reductions. However, the absolute loss in biomass was greatest in the gametophyte because starch was stored in much greater quantities than the sugars in all tissues, and most of the starch was in the gametophyte. The relative percentage reductions among the various seed tissues may not reflect the most influential facet of how this armored scale reduces seed quality. For example, the absolute decline in gametophyte starch may have a much greater impact on seed viability because this haploid tissue is constructed to support embryo growth and development. Germination studies will be required to determine the direct relationship between gametophyte starch and seedling germination and growth.

The outer sarcotesta tissue was the site of direct seed herbivory by the Hemiptera pest, and the size differential between sarcotesta depth and scale body size confirmed that the direct feeding was restricted to sarcotesta tissue. The role of direct photosynthesis by ovule and seed epidermis may also be a consequential factor in the carbon relations of maturing C. micronesica seeds. The exposed ovules and seeds of this species may remain green for more than 1 year of development (Marler and Dongol, 2016a), and A. yasumatsui infestations on these photosynthetic surfaces may impair the ability of seeds to act as their own carbon source. To our knowledge, no studies have been conducted to date on any Cycas species to quantify photosynthesis or respiration flux associated with the green surfaces of the naked seeds. Manipulative studies that physically exclude light energy from developing seeds may improve our understanding of the role of exposed Cycas seeds as carbon source. Carbon labeling studies would provide unambiguous evidence of the relative roles of concurrent carbon fixation by leaves, stored carbon in stems and roots, and direct seed photosynthesis in meeting the nonstructural carbohydrate needs of developing Cycas seeds.

Conservation implications.

The decline in seed resources that we have documented after direct A. yasumatsui infestation of seeds may partly explain why infested seeds exhibit reduced germination percentages and increased seedling mortality compared with uninfested seeds. These findings indicate that practitioners attempting to conserve genetic diversity of Cycas populations threatened by A. yasumatsui should collect a greater number of seeds from subpopulations that exhibit direct seed infestations than from subpopulations that have relatively clean seeds. This approach may enable a more even population structure of the resulting ex situ germplasm collections by taking into account the lower germination percentage and the greater seedling mortality from the heavily damaged subpopulations.

These results also underpin our assertions that A. yasumatsui continues to be the single primary threat to C. micronesica (Marler and Lindström, 2017). The enduring plant mortality is proceeding unabated due to failure to adequately establish efficacious biological control of the nonnative armored scale. For example, four of 12 unprotected trees in our first study and six of 20 trees in our second study died during the 16-month plan of work. Large sums of federal conservation funds are currently being expended to rescue unhealthy C. micronesica trees from military construction sites and transplant them to alternative recipient sites (Marler and Lindström, 2017). The primary threat of A. yasumatsui herbivory is not being addressed by any of these expensive conservation actions, and the transplanted trees will likely eventually die from this primary threat if this it continues to be ignored by conservation funding agencies. A redirection of these federal conservation funds toward mitigation of the primary threat would enable a more intelligent approach to species conservation, and if successful, the species could be delisted from the ESA.