This experiment was designed to investigate 2 factors: NAR genotype and time postharvest. The experiment used a low and high NAR genotype as described in [11] . Branches were monitored for 11 weeks postharvest, and each week was used as a separate time point for analysis. The experiment was replicated 10 times. Due to the destructive nature of some measurements 120 branches from each genotype were collected. Measurements and electron microscope observations were made on 20 branches each week (10 from each genotype). This process was repeated each subsequent week on 20 different branches.

All branches were collected from a clonal balsam fir orchard at the Tree Breeding Center, Department of Natural Resources, Debert, Nova Scotia, Canada (45˚25'N, 63˚28'W) in December. Each branch was cut from 2-year growth at 1.5 m aboveground. Collected branches were placed in a container with distilled water immediately after harvest to equilibrate water status. All branches were then transported to the lab.

All branches were provided a fresh cut 2.5 cm above the previous cut to ensure good water uptake. The fresh cut was made with the original cut end submerged in distilled water to reduce risk of introducing air into the xylem and disrupting water flow. Branches were weighed, placed in an amber bottle filled with 100 mL of distilled water, then the neck of each bottle was sealed with cotton wool to reduce direct water evaporation and provide added stability to a branch. All branches were randomly placed in a growth room at 21˚C constant temperature, 32% relative humidity, and 99 µmol∙m⁻²∙s⁻¹ light intensity.

Needle abscission was measured gravimetrically using dry mass [8] [29] . Each day the branches used to determine needles loss were subjected to a “finger run” test to dislodge loose needles, which was conducted by gently passing a branch between the index and middle finger 3 times. Fallen needles were collected and placed in an oven for 24 h at 90˚C. This process was repeated until the end of the 11th week when the experiment was terminated. Any needles still attached to a branch after the 11th week were removed from the branch, dried, then weighed so that percentage needle loss was expressed as a percentage of total needle dry mass. The length of time to abscise 1% of needles was defined as needle abscission commencement (NAC) [5] and was also reported in our experiment.

Water uptake was estimated gravimetrically. The mass of each bottle with water was weighed at the beginning of the experiment. Branches were removed from their bottles each week and the bottles were weighed again. Any difference in mass was estimated as water uptake. The rate of water uptake was calculated using Equation (1) and expressed in mL∙g⁻¹∙d⁻¹. In Equation (1), M_n is the mass of the flask on any given week, M_n+1 is the mass of the flask on the subsequent week, and M_b is fresh weight of the branch [29] . The difference in mass is also divided by a factor of 7 to convert the rate to daily uptake as opposed to weekly uptake.

WaterUptake = M n − M n + 1 7 × M b (1)

Dark-adapted chlorophyll fluorescence is used to assess maximum potential quantum and stress status of photosystem II [32] . Chlorophyll fluorescence in balsam fir needles was determined using an OS1p fluorometer (Opti-Sciences, Tyngsboro, MA) as described in [5] . Branches were placed in darkness for 15 min before fluorescence measurements, where a low intensity light pulse measured minimal fluorescence (F₀) and a saturating light pulse measured maximal fluorescence (F_m). Variable fluorescence (F_v) was calculated as the difference between F₀ and F_m. Dark adapted fluorescence was calculated as F_v/F_m.

Fully expanded needles from the previous year of growth were sampled immediately after harvest, 2-weeks postharvest, 4-weeks postharvest, and during abscission at 10-weeks postharvest. Needles were fixed by placing them in 2 mL vials of 2.5% glutaraldehyde then stored between 0 - 4˚C for at least 2 h. Needles from each genotype were placed in separate wells in a plastic 12-well plate tray at room temperature and then washed with tap water three times to remove the fixative with a plastic pipette. The needles were then fixed a second time with 2% osmium tetraoxide for 1 h at room temperature, after which needles were washed three more times with tap water. Needles were gradually dehydrated by coating and removing increasing concentrations of ethanol (25%, 50%, 70%, 80%, 90%, 95%, 100%). The highest concentration was used three times. The needles were then critical point dried using Leica EM CPD300 (Leica Microsystems Inc., Concord, ON). The dried needles were mounted on to SEM specimen stub with needle’s abaxial face facing up using conductive carbon cement. Three to four needles were mounted on each stub and coated with 12 nm of gold palladium alloy using a Lucia EM ACE200 (Leica Microsystems Inc., Concord, ON). Samples were viewed using a LEO 1455VP SEM (Carl Zeiss Canada Ltd., Toronto, ON).

Fully expanded needles from the previous year of growth were taken initially at the orchard and then each week from the lab for TEM. Needles were sampled at the same time of day (16:00 - 18:00). Needles were cut transversely and longitudinally and then fixed in 2.5% glutaraldehyde. Needles were fixed a second time in 1% osmium tetroxide and then dehydrated with acetone. Finally, needles were embedded in an epoxy resin for thin sectioning as described by [33] . Thin sections were observed in a JEM-1230 TEM (JEOL Canada Inc., Saint-Hubert, QC) equipped with an ORCA-HR high-resolution (2000 by 2000 pixels) digital camera (Hamamatsu Corp., Bridgewater, NJ).

This experiment was analyzed as a 2 × 12 factorial with 2 genotypes and 12 sample times. Statistical assumptions of normality and homogeneity were confirmed for quantitative analysis using Minitab 19 software (Minitab, LLC., Pennsylvania State College, PA). Independence was assured using different branches each week. Data were submitted to an analysis of variance using the general linear model feature in Minitab 19. Tukey’s multiple mean comparison was used to separate means at 5% significance. SEM and TEM photographs were described qualitatively.

The low NAR genotype had significantly higher needle abscission throughout the experiment than the high NAR genotype (Figure 1). Postharvest needle abscission began gradually in both genotypes during the first week, though the low NAR genotype lost 3.2 times more needles than the high NAR genotype. The low NAR genotype had lost over 1% of its total needles by week 2, which was 4.7 times more needles than the high NAR genotype. By comparison, the high NAR

genotype didn’t lose 1% of its total needles until week 5. Needle abscission dynamics for both genotypes followed a similar pattern, but the high NAR genotype was roughly 2 - 3 weeks slower than the low NAR genotype throughout the experiment.

The low NAR genotype had significantly lower chlorophyll fluorescence throughout the experiment than the high NAR genotype (Figure 2). The chlorophyll fluorescence of the high NAR genotype was 18% higher than the low NAR genotype immediately after harvest. Fluorescence increased in both genotypes during the first 3 weeks, then fluorescence decreased faster in the low NAR genotype. By week 7, chlorophyll fluorescence was 37% higher in the high NAR genotype than the low NAR genotype. Fluorescence started to significantly decrease in the high NAR genotype by week 9 but remained higher in high NAR genotype versus low NAR genotype. The high NAR genotype had 92% higher fluorescence than the low NAR genotype in the final week.

There was no significant difference in water uptake between genotypes in the first week (Figure 3). However, water uptake decreased by 79% in the low NAR genotype during the second week compared to only a 9% decrease in the high NAR genotype. The high NAR genotype maintained significantly higher water uptake than the low NAR genotype until week 5. There were no differences in water uptake between genotypes from week 5 until completion of the experiment.

Although there was relatively little fungus on balsam fir needles during harvest, there was noticeable more fungal contamination on low NAR genotypes versus high NAR genotypes during harvest (Figure 4). There was considerably more evidence of fungus in both genotypes by 2-weeks postharvest, though fungus was much more visible in low NAR genotypes (Figure 5). Fungus could be observed around stomata of low NAR genotypes at 151x magnification, which was further confirmed at 1010x magnification (Figure 5). Some stomata were also observed to have an unidentified waxy build up (Figure 6).

Both genotypes had easily observable fungus by 10-weeks postharvest (Figure 7). The high NAR genotype had several stomata clogged with fungus (Figure

7(a) and Figure 7(c)). However, virtually all stomata of low NAR genotypes were clogged and hyphae branching was more extensive than high NAR genotypes (Figure 7(b) and Figure 7(d)), It appeared that hyphae were trying to grow into the needle through the stomata in both clones. There were also bacteria, mold spores, pollen, and wax observed in both genotypes by 10-weeks postharvest (Figure 7). An examination of the abscission zone of a needle that had not yet been abscised by 10-weeks postharvest revealed the presence of many bacteria at the abscission zone (Figure 8).

Branches were preserved on site immediately during harvest and examined for chloroplasts and plastoglobuli (Figure 9). Chloroplasts were not significantly different in size between the two genotypes (Table 1). There were differences in both the size and number of plastoglobuli present. Plastoglobuli in high NAR genotypes were approximately 2x larger and occurred 3x more frequently than in low NAR genotypes. There was also some residual starch present in both genotypes, but it was not plentiful in either.

Chloroplasts in each genotype were actively producing starch grains 1-week postharvest (Figure 10). There appeared to be a clear separation of the stroma lamellae in the chloroplasts with larger starch deposits observed between the stroma lamellae. Thylakoids also appear to be swollen (Figure 10). Plastoglobuli were still present but were difficult to measure or quantify due to the increased number of starch grains. Chloroplasts were 46% smaller in both genotypes 1-week postharvest, but there were no differences in chloroplast size between genotypes.

The number of starch grains present in chloroplasts decreased in subsequent weeks postharvest. By 2 weeks postharvest it was more difficult to find chloroplasts or starch in the low NAR genotype (Figure 11). By 3 weeks postharvest, the

chloroplasts were over-filled with starch grains and chloroplast breakdown was evident (Figure 12). By 4 weeks postharvest chloroplasts in the high NAR genotype looked very much like they did at harvest with obvious plastoglobuli and little starch, which indicated inactivity. However, after 4 weeks postharvest the chloroplasts of low NAR genotypes displayed appreciable damage. Throughout

the next six weeks, it became increasingly difficult to find active chloroplasts in either genotype.

The final TEM images were taken 9-weeks postharvest and were particularly clear (Figure 13). Needles from the high NAR genotype had chloroplasts approximately 50% the size of chloroplasts at harvest. The chloroplasts in high NAR genotypes were also more globular and some starch grains remained visible (Figure 13(a)). However, the low NAR genotype displayed complete cellular and chloroplast disorganization 9 weeks postharvest (Figure 13(b)).

Low NAR genotypes abscised needle faster, had lower chlorophyll fluorescence, and decrease water uptake earlier than high NAR genotypes. The general trend in postharvest quality between contrasting genotypes is consistent with previous work [11] [29] [34] , though our study had a smaller difference in needle retention between low and high NAR trees. The difference in NAC in [34] was approximately 5 weeks, while our current study noted only a 3-week difference in NAC. There is significant variation in needle retention based on time of year, with cold acclimation often reducing the gap in postharvest needle abscission between genotypes [11] . Thus, it is possible that the time of harvest might help explain this discrepancy. Regardless, the purpose of studying postharvest characteristics was to confirm the difference between contrasting differences, which was successful in this study. From a practical perspective, this reaffirms that there is significant variance in needle retention characteristics of different balsam fir trees. Identification of trees with superior needle retention remains an important task.

It is well established that stomatal conductance rapidly decreases in balsam fir postharvest, which coincides with abscission commencement [12] [18] . There seems to be a degree of dehydration stress that can be observed through increases in abscisic acid and ethylene postharvest [8] [9] , but the contributing factors are not completely understood. One hypothesis was that bacterial accumulation in Christmas tree stand water was impeding uptake through the xylem, which was mitigated in part by antibacterial agents [12] . It can be suggested from our current study that stomatal dysfunction may also play a role in dehydration stress and needle abscission.

There were more fungal hyphae present on low NAR genotypes compared to high NAR genotypes. The exact nature of these fungi remains unknown, but vulnerability to harmful fungi could help explain why low NAR genotypes abscise needles so much earlier. Economic losses due to pathogenic fungi were once a concern in the balsam fir industry [35] , though more recent research has focused on fungal endophytes and balsam fir in New Brunswick, Canada. Endophytic fungi can have a mutual relationship with the plant, contributing to the defense of the host plant while not causing disease symptoms or additional distress [36] [37] . It is also possible for endophytic bacteria to modify certain phytohormones in host plants, such as abscisic acid [38] , which have been associated with postharvest abscission in balsam fir [8] .

It is also possible that we observed saprophytic fungi 11 weeks post-harvest. Saprophytic fungi, such as Coniophora puteana and Coriolus versicolor, are common in beech and pine [39] . Saprophytic fungi produce extracellular mucilaginous material (ECM) after colonizing on a tree, which accelerates senescence and decay [39] . The micro-photographs of ECM bear a resemblance to the “waxy build up” described in Figure 6 and Figure 7. It would be useful to identify the type of fungi present in future studies. Further, if fungi are contributing to postharvest abscission, then antifungal agents might delay abscission.

This study was the first to observe balsam fir needle chloroplasts postharvest using transmission electron microscopy. The chloroplast is a main target of abiotic stress, which often inhibits photosynthesis and degrades the photosynthetic apparatus [40] . Endogenous abscisic acid quickly increases in balsam fir postharvest, presumably to close stomata and conserve water [8] . Ethylene biosynthesis inevitably increases afterwards and promotes the synthesis of hydrolytic enzymes [9] [10] . Though not confirmed in balsam fir, it is likely that any increase in abiotic stress and ethylene evolution causes an imbalance in the redox state of plant cells as observed in other species [41] . A shift in the production of ROS without the complementary ability to sequester ROS will promote senescence [41] [42] , which is clearly demonstrated through a postharvest increase in abscission and a decrease in chlorophyll fluorescence in balsam fir. There are also several changes in balsam fir chloroplasts indicative of postharvest senescence.

There were changes in chloroplast starch postharvest. Starch is synthesized during the day and is indicative of active photosynthesis [43] . Starch may then be mobilized at night to help provide carbon for continued growth when not photosynthesizing [43] . However, starch is an emerging key molecule in abiotic stress response. Abiotic stress often limits photosynthesis and starch may be mobilized to provide energy, sugars, or other metabolites to help mitigate stress [44] . Little starch was present in balsam fir immediately after harvest suggesting limited photosynthesis, which was not surprising considering the cooler temperatures in December [45] [46] [47] [48] . Starch accumulated in balsam fir chloroplasts during the first week postharvest since plants were placed in warmer, lighted conditions. However, the low NAR and high NAR genotypes stopped producing starch in the second and fourth weeks postharvest, respectively. This lack of starch suggests that photosynthesis was inhibited and starch would likely have been mobilized from the chloroplast [44] .

There are several morphological changes to the chloroplast that are indicative of senescence. The three most relevant ultrastructural changes to describe the senescence-induced change of chloroplast into gerontoplasts are: 1) thylakoid membrane degradation, 2) increased plastoglobules number and size, and 3) disruption of the plastid envelope [41] [49] . All the above changes were observed in balsam fir chloroplasts postharvest but were more pronounced in the low NAR genotype. Further, chloroplasts decreased in size postharvest and began to transition from ellipsoid to circular morphology. Ethylene synthesis is increased in balsam fir postharvest [9] [10] and is also known to rapidly reduce the size of the chloroplasts [50] [51] . Further, ethylene causes the separation of thylakoid membrane layers due to degradation of granal and lamellar membranes [51] . Such a separation is evident in balsam fir as early as one week postharvest. There have been cases of gradual decreases in chloroplast size leading up to senescence, however, a fast decrease in one week is more specific to the action of ethylene [51] .

As mentioned earlier, abiotic stresses (e.g. dehydration) can induce ROS [41] [42] and chloroplasts are often one of the first organelles to be damaged during times of stress [52] . Degradation of chloroplasts accounts for postharvest decreases in chlorophyll fluorescence and helps explain potential differences in NAR genotypes observed in this study. Changes in lipids and fatty acids were linked to postharvest abscission in balsam [29] [34] . It was speculated that changes in the ratio of unsaturated and saturated fatty acids might explain genotypic differences in abscission [53] . Such changes in fatty acids would also help protect chloroplasts [53] , potentially explaining why chloroplasts were protected to some degree in high NAR genotypes

Chloroplasts became especially difficult to detect several weeks postharvest in the low NAR genotype. This could indicate the start of autophagy to remove damaged organelles [54] . Damaged cells may also induce production of ubiquitin ligase, an enzyme that signals for the degradation of ROS producing chloroplasts [55] [56] . Another possibility for difficulty in detecting chloroplasts is that often the osmophilicity of tissues changes when they are damaged or senescing, which means that osmium tetroxide may have been less effective when preparing samples for TEM. However, it seems less likely that deficiencies in osmium tetroxide staining would be a problem since the same protocol was used for all slide preparation and weekly samples of each genotype were prepared at the same time.