Thick bark can protect trees from a severe ambrosia beetle attack

Study Site and Study Organisms

The study was conducted in the Tijuana River Valley, a small (14.6 square kilometer) coastal floodplain in San Diego County, California, at the end of a 4,480 square kilometer, binational watershed (Safran et al., 2017). The river is an intermittent stream that typically flows strongly in winter and spring and is mostly dry in summer (Boland, 2014). Because of frequent cross-border inputs of sewage, it is one of the most polluted rivers in the state (Gersberg, Daft & Yorkey, 2004; Boland & Woodward, 2019). The riparian forests in the valley are preserved within three adjoining open space parks, and two willow species, S. gooddingii and S. lasiolepis Benth., are numerically and structurally dominant (Boland, 2014). In 2015, when the KSHB was first detected in the valley, the riparian habitats were divided into 29 units for census purposes such that each unit was relatively homogenous in terms of species composition, willow age and willow density (Boland, 2016), and those units were retained for this study. The units can be grouped into Wet Forests, where the trees are flooded each winter, and Dry Forests, where the trees rarely receive river flows. The Dry Forests are dry because the river has migrated away from them. All of the Wet Forests were initially heavily damaged by the KSHB (Boland, 2016) but since 2016 they have substantially recovered and in 2020 they were similar to their pre-KSHB stature (Boland & Uyeda, 2020; Boland et al., 2020).

Female KSHB beetles bore through the bark and create their galleries inside the sapwood (xylem) of live trees. They carry with them the symbiotic fungi, Fusarium kuroshium and Graphium kuroshium, which they farm in the gallery walls and consume (Na et al., 2018). The closely related and invasive Polyphagus Shot Hole Borer, Euwallacea whitfordiodendrus (Schedl), also attacks live trees in southern California, and the damage caused by it and the KSHB is being called Fusarium Dieback (Umeda, Eskalen & Paine, 2016; California Department of Fish Wildlife, 2020). Specimens from the Tijuana River Valley were collected and identified as KSHB by Akif Eskalen at University of California Riverside and are stored in their collection. The County of San Diego Department of Parks and Recreation and the US Fish and Wildlife Service allowed access to their properties via permits ROE08.04.16 and 19001TJS, respectively.

Bark samples

Bark samples were collected from 27 S. gooddingii trees in 2016–17, during the peak of the KSHB infestation in the valley (Boland & Uyeda, 2020). The sampled trees ranged from heavily infested and nearly dead to lightly infested and seemingly unaffected. They were from nine forest units where the stands were 0 m to 520 m from the river and five to 35 years old (Boland, 2016; Boland & Woodward, 2019). Of the nine forest units, five were Wet Forests (Units 2, 3, 9, 12 and 13) and four were Dry Forests (Units 15, 17, 19 and 21). In each unit, three infested trees were randomly chosen and, at each tree, the trunk diameter was measured and a bark sample was collected at breast height (1.37 m). Each bark sample was approximately 11 × 20 cm and cut using a hand saw, hammer, and chisel. The samples included inner and outer bark layers and no xylem. The scar was sprayed with TreeKote Tree Wound Dressing to reduce the risk of infection, and tools were cleaned between samples with a disinfectant wet wipe.

In the lab and within hours of collection, a 10 × 15 cm grid was drawn on the inner surface of each bark sample using a ruler and permanent marker, and the following counts and measurements were taken under a 3.5x desk magnifier: (1) the number of KSHB holes in the grid (# holes per 150 cm²); (2) the bark thickness at each hole (n = 0 to 113 holes per sample); and (3) the bark thickness at each grid line intersection (n = 150 points per sample). In total, 914 bark thickness measurements were made at KSHB holes and 4,050 bark thickness measurements were made at grid line intersections; these represented the thicknesses used by the KSHB and the thicknesses available to the KSHB, respectively. All bark thicknesses were measured using digital vernier calipers (Vinca DCLA-0605) fitted with extensions (Anytime Tools 4 Points Caliper Extension Set for Deep Measuring), allowing measurements at all points on the sample. Extension tips were two mm in diameter, small enough to measure in narrow furrows but not so small as to extend into the KSHB holes, which are approximately one mm in diameter. All bark thickness measurements were recorded to three decimal places (0.000 cm) and are presented to two decimal places. Precision was quantified by measuring the 150 grid line intersections on one bark sample twice and comparing the two sets of measurements; the Root Mean Square Error of the paired measurements was 0.07, the difference in median thickness was 0.014 cm, and the difference in mean thickness was 0.002 cm, indicating good precision.

Our use of relatively large bark samples, which included several of the vertical ridges and furrows characteristic of S. gooddingii bark, allowed us to document the range of bark thicknesses available to and used by the KSHB and to accurately count KSHB holes from the inner surface of the bark. Holes are often visible on the outer surface of the bark but are more reliably counted on the inner surface when bark is textured. Lenticels are areas of loosely packed cells in bark that allow for gas exchange, and potentially could be vulnerable spots for entry of boring insects (Rosner & Führer, 2002); however, lenticels in S. gooddingii are inconspicuous and small relative to KSHB holes, so we were unable to ascertain the use of lenticels as points of entry by KSHB. Throughout this paper, the term ‘holes’ refers to the bore holes that KSHB make straight through the bark into and out of their galleries in the xylem; the term ‘furrows’ refers to the cracks, crevices, grooves or fissures where bark is relatively thin; and the term ‘ridges’ refers to the ribs or spurs where bark is relatively thick (Vaucher & Eckenwalder, 2003; Wojtech, 2011).

Bark thickness influences KSHB attack densities and attack locations (H1)

To test this hypothesis, the bark sample data were used in three ways. First, the relationship between the density of KSHB holes in a sample and the median bark thickness of the sample was examined using exponential regression. Because there were some hole densities of zero, a value of 0.1 holes was added to all densities to obtain the exponential best-fit line. Second, to test whether bark thickness influenced KSHB attack locations, the mean of all the bark thicknesses used (n = 914) was compared to the mean of all the bark thicknesses available (n = 4,050) using a t-test. Third, to test whether KSHB showed a preference for certain bark thicknesses, the similarity between bark thicknesses used and available was determined for each bark sample. A measure of similarity, the proportional similarity index (PSI; Zaret & Smith, 1984), was calculated for each bark sample. The PSI is:

PSI = ∑ min (p_i, q_i)

where p_i and q_i are the proportion available (p) and proportion used (q) for each thickness increment (i), using 0.1 cm increments over the range <0.2 to 2.0 cm. In short, the PSI was a measure of how similar the frequency distributions of bark thicknesses available and used were for each bark sample. A high PSI meant a high similarity, or high overlap, in the frequency distributions and indicated the random choice of bark thickness by the KSHB. A low PSI meant a low similarity, or low overlap, in the frequency distributions and indicated a strong deviation from random choice of bark thickness by the KSHB. Exponential regression was used to test for significant correlation between the PSI and the median bark thickness.

Bark thickness influences KSHB impacts (H2)

To test this hypothesis, the bark sample data were used with other data in four ways. First, the relationship of bark thickness to trunk diameter was examined to confirm that the expected relationship of bark thickness increasing with tree size (Rosell, 2016) occurs in S. gooddingii. Second, the relationship of KSHB hole density to trunk diameter was examined. Because there were some hole densities of zero, a value of 0.1 holes was added to all densities to obtain the exponential best-fit line. Third, the relationship of bark thickness to willow mortality rate was examined. Willow mortality rates for each unit were the maximum percent mortality obtained from three annual surveys of living and dead willows conducted in each unit between 2015 and 2018 (Boland, 2019). Fourth, to better understand the marked spatial pattern in the KSHB impacts in the valley, i.e., that willow mortality rates were significantly higher near the main river channel (Boland & Woodward, 2019), we compared the mean bark thicknesses of the samples from the Wet and Dry Forest units using a t-test.

Comparison of two fallen trees

Two S. gooddingii trees that had fallen during 2017 provided an opportunity to estimate the total impact of KSHB on trees of different sizes. Both trees had been infested by KSHB while they were alive and standing, both had been blown down by strong winds, and both were examined soon after they fell. The smaller tree had broken at 1.4 m on the trunk (wind-snapped) due to extensive KSHB tunneling that undermined the structural integrity of the wood, whereas the larger tree had been uprooted (wind-thrown) with its trunk intact. The smaller tree (11 m tall and 10.8 cm diameter at breast height) was in a Wet Forest (Unit 2) in the current river bed, and the larger tree (23 m tall and 40.7 cm diameter at breast height) was in a Dry Forest 60 m from the current river bed (Unit 19; unit locations in Boland, 2016). Along the full trunk length of each tree, 1 m sections were marked and three measurements were taken within each section: maximum bark thickness, density of KSHB holes, and trunk circumference. Maximum bark thickness was measured in a bark sample (approximately 8 × 5 cm) cut with a hand-saw, hammer, and chisel such that the sample included inner and outer bark layers and no xylem; maximum thickness was measured using a pair of digital vernier calipers (Vinca DCLA-0605). KSHB holes were counted within a flexible quadrat (20 × 2 cm = 40 cm²) placed parallel to the axis of the trunk, and trunk circumference was measured with a tape.

To estimate the KSHB impact on the two trees, the volume of xylem damaged by KSHB and its associated fungi was estimated for each section of tree trunk. This estimate was done by assuming that each KSHB hole in the bark represented a 2.5 cm³ impact to the xylem. For each 1 m section of trunk, the total volume was calculated as a cylinder from the circumference measurement, the total number of KSHB holes was extrapolated from the density in the quadrat, the KSHB-impacted volume was calculated from the total number of holes multiplied by 2.5 cm³, and the KSHB-impacted volume was expressed as a percent of the total volume of the section. The assumption of a 2.5 cm³ impact per hole is based on measurements of many KSHB tunnel lengths (113 mm tunnel length per hole on average) and many fungal-induced cell deformities visible around KSHB tunnels (ellipse with a major radius of 4.7 mm and a minor radius of 1.5 mm on average; Boland unpublished data). This is a relatively conservative estimate of impact because others have found more extensive fungal impacts around the tunnels of other ambrosia beetles (Takahashi, Matsushita & Hogetsu, 2010).

Bark samples

Overall, measurements of the bark thicknesses available ranged from 0.12 cm to 2.74 cm (n = 4, 050) and measurements of the bark thicknesses used ranged from 0.12 cm to 1.07 cm (n = 914). Among the individual bark samples, median thicknesses available ranged from 0.26 cm to 1.89 cm (n = 27) and KSHB hole densities ranged from 0 to 113 holes per 150 cm² (Table 1). Seven samples had no holes, indicating that the trees from which they were taken were only lightly infested and holes were rare.

Bark thickness influences KSHB attack densities and attack locations (H1)

The results supported the hypothesis that bark thickness influences KSHB attack densities and attack locations. First, hole density and median bark thickness were significantly and negatively correlated, indicating that thicker bark samples had significantly fewer holes (r² = 0.676, n = 27, p < 0.01; Fig. 1). Second, the mean bark thickness used (0.52 ± 0.20 cm; n = 914) was significantly less than the mean bark thickness available (0.97 ± 0.54 cm; n = 4,050; t-test, t = 25.06, p < 0.01; Fig. 2), indicating that the KSHB did not use bark thicknesses in proportion to those available. KSHB bored into relatively thin bark (<0.8 cm) at a higher frequently than its availability and bored into relatively thick bark (>0.8 cm) at a lower frequently than its availability. No KSHB holes were in bark thicker than 1.07 cm, even though thicker bark was abundantly available. Third, the PSI declined significantly with increasing bark thickness, indicating that similarity between the thicknesses used and available decreased with bark thickness (r² = 0.662, n = 27, p < 0.01; Fig. 3). In summary, bark samples that were generally thin had many KSHB holes and these holes were spread in a random distribution (high similarity between thicknesses used and available), whereas bark samples that were generally thick had few KSHB holes and these holes were found almost exclusively in furrows in a non-random distribution (low similarity between thicknesses used and available).

Two of the 27 bark samples are contrasted here to illustrate the above results (Figs. 4, 5 and 6). An external view shows that the thinner sample had many KSHB holes dispersed over the bark surface, whereas the thicker sample had fewer holes confined to furrows (red pins in Fig. 4). A profile view shows the same dispersion in cross section, i.e., the thinner sample had many KSHB holes dispersed over the bark surface, and the thicker sample had fewer holes confined to furrows, where bark was thinnest (red dots in Fig. 5). The similarity between the frequency distributions of thicknesses available and used was high in the thinner sample (PSI = 94%) indicating the random distribution of KSHB holes, whereas the similarity was low in the thicker sample (PSI = 9%) indicating the non-random distribution of holes and site selection on the part of the KSHB (Fig. 6). As these results show and as these two samples illustrate, bark thickness influenced both KSHB attack densities and attack locations, i.e., the KSHB bored abundantly through thin bark and avoided boring through thick bark.

Bark thickness influences KSHB impacts (H2)

The results also supported the hypothesis that bark thickness influences KSHB impacts. First, trunk diameters of the 27 sampled trees ranged from 8.9 to 39.2 cm (Table 1), and, as expected, all measures of bark thickness were positively correlated with trunk diameter, i.e., median bark thickness increased significantly with increasing diameter (r² = 0.686, n = 27, p < 0.01), as did maximum (r² = 0.663, n = 27, p < 0.01) and minimum (r² = 0.400, n = 27, p < 0.01) bark thicknesses (Fig. 7). Second, the density of KSHB holes was negatively correlated with trunk diameter, i.e., hole density decreased significantly with increasing trunk diameter (r² = 0.409, n = 27, p < 0.01; Fig. 8). Third, willow tree mortality rates in the nine units from which the samples were collected ranged from 0 to 97%, and those mortality rates were negatively correlated with bark thickness, i.e., mortality rates decreased significantly as median bark thickness increased (r² = 0.544, n = 27, p < 0.01; Fig. 9). In summary, S. gooddingii trees with thick bark had larger diameters, fewer holes per unit area, and lower mortality rates, all consistent with the hypothesis that bark thickness influences KSHB impacts such that KSHB causes more damage to smaller, thinner-barked trees than to larger, thicker-barked ones.

Finally, the mean bark thickness of samples collected from the Wet Forest units was 0.67 cm (± 0.19 cm; n = 15) whereas the mean bark thickness of samples collected from Dry Forest units was 1.39 cm (± 0.28 cm; n = 12). These bark thicknesses were significantly different (t-test, t = 7.93, p < 0.01), indicating that the more vulnerable, thin-barked trees were significantly more common in the Wet Forests.

Comparison of two fallen trees

In both fallen trees, the maximum bark thickness was greatest on the lower trunk and declined upward (Fig. 10A), the density of KSHB holes was greatest on the lower trunk and declined upwards (Fig. 10B), and the vertical range of infestation was from ground level to about 8 m. Over this range of infestation, the larger tree had considerably thicker bark than the smaller tree (Fig. 10A). As for impact, the estimates of wood volume impacted by the KSHB were strikingly different in the two trees. In the larger tree, the estimated impact was less than 12% in all sections, whereas in the smaller tree the estimated impact was well over 50% in four of eleven sections and up to 96% in one section (Fig. 10C). The low density of holes in the larger, thicker-barked tree therefore led to negligible impacts, whereas the high density of holes in the smaller, thinner-barked tree led to the near destruction of its xylem within large sections of its trunk. In summary, this comparison of the two fallen trees shows why a KSHB infestation does not kill a thick-barked tree but can kill a thin-barked tree.