Investigating the Relationship between Nitrate, Total Dissolved Nitrogen, and Phosphate with Abundance of Pathogenic Vibrios and Harmful Algal Blooms in Rehoboth Bay, Delaware

ABSTRACT

Vibrio spp. and phytoplankton are naturally abundant in marine environments. Recent studies have suggested that the co-occurrence of phytoplankton and the pathogenic bacterium Vibrio parahaemolyticus is due to shared ecological factors, such as nutrient requirements. We compared these communities at two locations in the Delaware Inland Bays, representing a site with high anthropogenic inputs (Torquay Canal) and a less developed area (Sloan Cove). In 2017 to 2018, using light microscopy, we were able to identify the presence of many bloom-forming algal species, such as Karlodinium veneficum, Dinophysis acuminata, Heterosigma akashiwo, and Chattonella subsalsa. Dinoflagellate biomass was higher at Torquay Canal than that at Sloan Cove. D. acuminata and Chloromorum toxicum were found only at Torquay Canal and were not observed in Sloan Cove. Most probable number real-time PCR revealed V. parahaemolyticus and Vibrio vulnificus in environmental samples. The abundance of vibrios and their virulence genes varied between sites, with a significant association between total dissolved nitrogen (TDN), PO₄⁻, total dissolved phosphorus (TDP), and pathogenic markers. A generalized linear model revealed that principal component 1 of environmental factors (temperature, dissolved oxygen, salinity, TDN, PO₄⁻, TDP, NO₃:NO₂, NO₂⁻, and NH₄⁺) was the best at detecting total (tlh+) V. parahaemolyticus, suggesting that they are the prime drivers for the growth and distribution of pathogenic Vibrio spp.

IMPORTANCE Vibrio-associated illnesses have been expanding globally over the past several decades (A. Newton, M. Kendall, D. J. Vugia, O. L. Henao, and B. E. Mahon, Clin Infect Dis 54:S391–S395, 2012, https://doi.org/10.1093/cid/cis243). Many studies have linked this expansion with an increase in global temperature (J. Martinez-Urtaza, B. C. John, J. Trinanes, and A. DePaola, Food Res Int 43:10, 2010, https://doi.org/10.1016/j.foodres.2010.04.001; L. Vezzulli, R. R. Colwell, and C. Pruzzo, Microb Ecol 65:817–825, 2013, https://doi.org/10.1007/s00248-012-0163-2; R. N. Paranjpye, W. B. Nilsson, M. Liermann, and E. D. Hilborn, FEMS Microbiol Ecol 91:fiv121, 2015, https://doi.org/10.1093/femsec/fiv121). Temperature and salinity are the two major factors affecting the distribution of Vibrio spp. (D. Ceccarelli and R. R. Colwell, Front Microbiol 5:256, 2014, https://doi.org/10.3389/fmicb.2014.00256). However, Vibrio sp. abundance can also be affected by nutrient load and marine plankton blooms (V. J. McKenzie and A. R. Townsend, EcoHealth 4:384–396, 2007; L. Vezzulli, C. Pruzzo, A. Huq, and R. R. Colwell, Environ Microbiol Rep 2:27–33, 2010, https://doi.org/10.1111/j.1758-2229.2009.00128.x; S. Liu, Z. Jiang, Y. Deng, Y. Wu, J. Zhang, et al. Microbiologyopen 7:e00600, 2018, https://doi.org/10.1002/mbo3.600). The expansion of Vibrio spp. in marine environments calls for a deeper understanding of the biotic and abiotic factors that play a role in their abundance. We observed that pathogenic Vibrio spp. were most abundant in areas that favor the proliferation of harmful algal bloom (HAB) species. These results can inform managers, researchers, and oyster growers on factors that can influence the growth and distribution of pathogenic Vibrio spp. in the Delaware Inland Bays.

INTRODUCTION

The Delaware Inland Bays (DIBs) are a collection of saltmarshes, tidal flats, oyster reefs, saltwater creeks, and shallow open waters. The DIBs are located in the mid-Atlantic region of the United States, near Rehoboth Beach and Fenwick Island, Delaware. The watershed consists of ~750 square kilometers of land area draining into ~90 square kilometers of bay and tributaries (1). The intensity and duration of harmful algal bloom (HAB) events have been increasing in the United States (2, 3), and a similar pattern has been observed in the DIBs (4–6). These issues have led to concerns regarding potential human exposure to toxins produced by HAB species.

HABs, such as Karlodinium veneficum and Dinophysis sp., have been detected in tributaries near Little Assawoman Bay, Delaware (6). Dinophysis spp. are known to cause diarrhetic shellfish poisoning (DSP) via the production of okadaic acid (OA) and its derivatives, the dinophyistoxins (DTXs), which are lipophilic toxins that accumulate in the fatty tissue of shellfish (7). Blooms of raphidophytes have also been studied extensively in the DIBs and have been responsible for fish mortalities (4, 8). Heterosigma akashiwo has been detected regularly in the DIB, and it has the potential to cause sublethal impacts on the eastern oyster (8).

Vibrio spp. are naturally occurring Gram-negative marine bacteria that are found in diverse habitats ranging from coastal to open waters (9). They can survive as free-living organisms or attached to organic particles and biofilms (10). They are able to degrade polymeric substrates, such as chitin, plant/algal polysaccharides, and plastic waste (10). Vibrio can colonize and degrade particulate matter and consequently play an important role in chemical transformations, which contributes to cycling of carbon and other nutrients (11). Studies have suggested that organic nutrients that stimulate algal blooms of Gymnodinium, Dinophysis, and Ceratium may be the most significant factor driving interrelationships between Vibrio spp. and phytoplankton (12, 13). In the DIB, Vibrio spp. have been correlated with particulate matter of >20 μm, which includes phytoplankton, such as diatoms and raphidophytes (14).

A combined abiotic and biotic analysis was performed at two sites in the DIBs, as follows: one, which is known to be impacted negatively by anthropogenic factors (Torquay Canal), and another that is in proximity to a proposed aquaculture site (Sloan Cove). At each location, we (i) analyzed water quality parameters, such as temperature, dissolved oxygen, salinity, and nutrients; (ii) identified and enumerated the HAB species present at both locations; and (iii) examined the levels of total and pathogenic Vibrio parahaemolyticus and Vibrio vulnificus from collected water. The goal of the study was to determine the most influential factors in the proliferation of two potentially pathogenic Vibrio species accounting for environmental factors that may simultaneously influence the development of algal blooms, so as to inform management decisions for aquaculture efforts.

RESULTS

Comparison of environmental parameters between sampling sites, namely, temperature, salinity, dissolved oxygen, and nutrients.

Abiotic parameters were measured at two sites in the DIBs (Fig. 1) to establish the relative water quality between sites. On average, the water temperature at Torquay Canal was 24.4°C (range of 12.3 to 31.4°C), and at Sloan Cove, it was 23.5°C (10.4 to 23.5°C). The average salinity at Torquay Canal was 25.6 g/kg (range of 18.1 to 25.6 g/kg), and at Sloan Cove, it was 27.9 g/kg (12.2 to 30.4 g/kg). However, a Mann-Whitney rank sum test revealed that there was not a significant difference in temperature or salinity between sites (see Fig. S1 in the supplemental material).

FIG 1.

Location of sampling sites in the Delaware Inland Bays. 1, Torquay Canal; 2, Sloan Cove; (The maps were created with ggmaps on R studio [45]).

Water quality at Torquay Canal was degraded compared with that of Sloan Cove (Table 1), with dissolved oxygen averaging 3.9 mg mL⁻¹ and 6.3 mg mL⁻¹, respectively. According to a Mann-Whitney rank sum test, there were significant differences in total dissolved phosphorus (TDP), PO₄⁻³, and total dissolved nitrogen (TDN) between sites (P < 0.05). TDP at Torquay Canal averaged 3.48 μM compared with that of 1.46 μM at Sloan Cove, and PO₄⁻³ levels averaged 1.68 μM at Torquay Canal compared with that of 0.81 μM at Sloan Cove, representing an ~2-fold higher concentration of these important phosphorus constituents. There was also a 1.4-fold elevation of TDN at Torquay Canal, which averaged 48.22 μM compared with that of 33.75 μM at Sloan Cove.

TABLE 1.

Average distribution of environmental parameters in Rehoboth Bay water samples from 2017 to 2018

Parameter	Values by location				P valuea
	Sloan Cove		Torquay Canal
	Avg	Min/max	Avg	Min/max
Dinoflagellate biomassb	8.60 × 101	0/7.20 × 103	4.60 × 103	0/2.80 × 103	0.04
K. veneficum biomassb	1.40	0/6.00 × 103	6.50	0/4.90	0.55
Tempc	23.50	10.40/31.60	24.40	12.30/31.40	0.77
Salinityd	27.90	12.20/30.40	25.60	18.10/30.90	1.00 × 10−3
Dissolved oxygene	6.40	3.70/8.80	3.90	1.26/6.19	1.10 × 10−5
NH4+f	4.74	0.69/29.30	6.18	0.54/28.20	0.14
NO3−:NO2−f	1.35	0.32/4.99	1.80	0.18/6.06	0.35
NO2−f	0.40	0.15/1.41	0.43	0.13/1.26	0.16
TDNf	33.75	13.1/169.00	48.22	15.1/138.00	1.68 × 104
PO43−f	0.81	0.31/1.68	1.68	0.29/7.34	8.99 × 10−4
TDPf	1.46	0.74/3.18	3.48	0.51/17.50	1.85 × 10−5

^aBoldface entries indicate significant P values.

^bμgC L⁻¹.

^c°C.

^dg/kg.

^emg·L⁻¹.

^fμM.

Harmful algal community dynamics.

As a means of assessing relationships between nutrients and HABs, phytoplankton analyses were performed at each site. The monthly distributions of known HAB species at Torquay Canal and Sloan Cove are shown in Fig. 2 and 3. At Torquay Canal, there was a frequent detection of phytoplankton blooms, including many HAB species, such as Dinophysis acuminate, Chloromorum toxicum, Karlodinium veneficum, and Heterosigma akashiwo. Gymnodinium spp. (combination of Gymnodinium auerolum and Gymnodinium instratium), Scrippsiella trochoidea, and Prorocentrum minimum were also identified routinely. At Sloan Cove, we observed many of the same phytoplankton, although potential toxin-forming species were of lesser prevalence and intensity and Dinophysis acuminata and Chloromorum toxicum were not detected.

FIG 2.

Average distribution of dinoflagellates and raphidophytes in Torquay Canal. The lack of a bar signifies no detection of dinoflagellates or raphidophytes.

FIG 3.

Average distribution of dinoflagellates and raphidophytes in Sloan Cove. The lack of a bar signifies no detection of dinoflagellates or raphidophytes.

According to the Mann-Whitney rank sum test, there was no significant difference in HAB concentrations between sites and years. However, there was a significant difference in dinoflagellate biomass between sites, with a P value of 0.02 (Fig. 4). The total dinoflagellate biomass was 1.4 × 10⁵ μgC·L⁻¹ (maximum observed in June 2018) at Torquay Canal and 3.7 × 10⁴ μgC·L⁻¹ at Sloan Cove (maximum observed in August 2018). At Torquay Canal, the total K. veneficum biomass was 1.6 × 10 ⁰² μgC·L⁻¹ (maximum observed in May 2017), and it was 133.6 μgC·L⁻¹ at Sloan Cove (maximum observed in May 2018). A Mann-Whitney rank sum test found no significant difference in K. veneficum biomass (Table 1).

FIG 4.

Average biomass of dinoflagellate collected from Sloan Cove and Torquay Canal in 2017 to 2018. (a) Dinoflagellates of >20 μm in size, including D. acuminata, G. aureolum, G. instratium, S. trochoidea, and P. minimum. (b) Dinoflagellates of <20 μm in size, including K. veneficum.

A principal-component analysis (PCA) was also performed to reveal overall relationships within phytoplankton species and environmental factors and between phytoplankton groups and environmental variables. An analysis of Torquay Canal and Sloan Cove environmental data showed an association between NO₃:NO₂, NO₂⁻, and NH₄⁺ and an association between TDN, TDP, and PO₄³⁻ (Fig. 5). Additionally, there was an inverse relationship between low salinity and higher nitrogen constituents. Phytoplankton data showed specific groupings between the following: Gymnodinium aureolum, K. veneficum, H. akashiwo, and Mesodinium rubrum; C. subsalsa, C. toxicum, P. minimum, and cryptophytes; and D. acuminata and G. instratium (Fig. 5). Nutrient and phytoplankton PCA showed that there was an association between cryptophytes, TDP, TDN, and PO₄³⁻ concentrations. K. veneficum, H. akashiwo, and G. aureolum also were associated with M. rubrum concentrations (see Fig. S2 in the supplemental material).

FIG 5.

Principal-component analysis of environmental factors and phytoplankton species collected from April to October 2017 to 2018. An abbreviation description is as follows: (a) Saln, salinity; Temp, temperature; DO, dissolved oxygen; (b) P.min, P. minimum; D.acu, D. acuminata; G.ins, G. instratium; G.aur, G. aureloum; S.tro, S. trochoidea; C.sub, C. subsalsa; C.tox, C. toxicum; H.aka, H. akashiwo; Cry, Cryptomonads; K.ven, K. veneficum; and M.rub, M. rubrum.

V. parahaemolyticus levels in water.

Thermolabile hemolysin (tlh+) V. parahaemolyticus was detected in 20/26 (77%) of water samples from Torquay Canal and in 22/29 (76%) water samples from Sloan Cove (Table 2 and 3). Tlh+ V. parahaemolyticus levels in Torquay Canal and Sloan Cove ranged from 0.97 to 3.04 log most probable number (MPN) mL⁻¹. Pathogenic thermostable direct hemolysin-positive (tdh+) V. parahaemolyticus was detected in 12/26 (46%) of water samples from Torquay Canal and 5/29 (17%) of water samples from Sloan Cove. Tdh+ V. parahaemolyticus concentrations in Torquay Canal and Sloan Cove ranged from 0.47 to 1.63 log MPN mL⁻¹. According to a Fisher exact test, Torquay Canal had a statistically significant higher number of tdh+ V. parahaemolyticus than Sloan Cove with a P value of 0.01.

TABLE 2.

Distribution of Vibrio spp., bloom species, and abiotic parameters in Torquay Canal

Variable	No. positive/total	Percent positive	Min	Max	Avg
V. parahaemolyticus (tlh+)a	20/26	77	0.97	3.04	1.78
V. parahaemolyticus (tdh+)a	12/26	46	0.79	1.36	0.54
V. parahaemolyticus (trh+)a	9/26	35	0.48	1.36	0.388
V. vulnificus (vvhA+)a	15/26	58	0.50	2.40	1.16
V. vulnificus (vcgC+)a	12/26	46	0.48	1.36	0.49
Dinophysis acuminata b	4/26	15	1.04 × 103	1.04 × 105	4.50 × 103
Gymnodinium spp.b	14/26	54	1.04 × 103	2.14 × 105	1.28 × 104
Karlodinium veneficum b	4/26	15	2.08 × 103	5.20 × 104	2.99 × 103
Scrippsiella trochoidea b	12/26	46	1.04 × 103	6.86 × 104	5.16 × 103
Prorocentrum minimum b	11/26	42	1.04 × 103	6.86 × 104	5.43 × 103
Heterosigma akashiwo b	5/26	19	1.04 × 103	1.99 × 104	8.52 × 104
Chloromorum toxicum b	4/26	15	3.36 × 102	5.65 × 105	1.42 × 105
Tempc	nad		12.3	31.4	24.4
Salinitye	na		18.1	30.9	25.6
Dissolved oxygenf	na		1.26	6.19	3.9

^alog MPN mL⁻¹.

^bCells L⁻¹.

^c°C.

^dna, not applicable.

^eg/kg.

^fmg L⁻¹.

TABLE 3.

Distribution of Vibrio spp., bloom species, and abiotic parameters in Sloan Cove

Variable	No. positive/total	Percent positive	Min	Max	Avg
V. parahaemolyticus (tlh+)a	22/29	76	1.36	3.04	1.73
V. parahaemolyticus (tdh+)a	5/29	17	0.47	1.63	0.20
V. parahaemolyticus (trh+)a	5/29	17	0.47	1.36	0.15
V. vulnificus (vvhA+)a	16/29	55	0.97	4.64	0.96
V. vulnificus (vcgC+)a	6/29	21	0.47	1.36	0.21
Dinophysis acuminata b	0/29	0	na	na	na
Gymnodinium spp.b	9/29	31	1.04 × 103	7.28 × 103	2.53 × 103
Karlodinium veneficum b	7/29	24	1.04 × 103	5.02 × 103	1.11 × 103
Scrippsiella trochoidea b	13/29	45	1.04 × 103	8.32 × 103	1.65 × 104
Prorocentrum minimum b	7/29	24	1.04 × 103	7.28 × 103	1.42 × 103
Heterosigma akashiwo b	3/29	10	2.08 × 103	6.24 × 103	3.47 × 102
Chloromorum toxicum b	0/29	0	na	na	na
Tempc	na		10.4	31.6	23.5
Salinityd	na		12.2	30.4	27.9
Dissolved oxygene	na		3.7	8.8	6.4

^alog MPN mL⁻¹.

^blog cells L⁻¹.

^c°C.

^dg/kg.

^emg L⁻¹.

Pathogenic thermostable related hemolysin-positive (trh+) V. parahaemolyticus was detected in 9/26 (35%) water samples from Torquay Canal and 5/29 (17%) water samples from Sloan Cove. Trh+ V. parahaemolyticus levels in Torquay Canal and Sloan Cove ranged from 0.47 to 1.36 MPN mL⁻¹. As with tdh+ V. parahaemolyticus, Torquay Canal was significantly higher in the number of trh+-positive V. parahaemolyticus compared with Sloan Cove, with a P value of 0.02.

V. vulnificus levels in water.

A V. vulnificus hemolysin A gene (VvhA+) was detected in 15/26 (58%) water samples collected from Torquay Canal and 16/29 (55%) from Sloan Cove. VvhA+ V. vulnificus levels at Torquay Canal and Sloan Cove ranged from 0.50 to 4.64 log MPNml⁻¹. A Vibrio-correlated clinical gene (VcgC+) V. vulnificus was detected in 12/26 (46%) water samples from Torquay Canal and 6/29 (21%) from Sloan Cove. VcgC+ V. vulnificus levels in Torquay Canal and Sloan Cove water samples ranged from 0.47 to 1.36 MPN mL⁻¹. VcgC+ V. vulnificus-positive samples were significantly higher at Torquay Canal than those at Sloan Cove with a P value of 0.03.

Comparison between years of V. parahaemolyticus and V. vulnificus.

We also compared the distribution of V. parahaemolyticus and V. vulnificus in water between 2017 and 2018, pooling data from each site. In 2017, samples were collected only between July and October; thus, we pooled only data for these months for a comparative analysis. In 2017, the highest levels of the tlh+ and tdh+ V. parahaemolyticus were observed (Table 4), and there was a significant difference in the tlh+ and tdh+ levels between these years (P values 0.003 and 0.038, respectively). The highest concentration of vvhA+ V. vulnificus in water samples was observed in August 2017, and the highest concentration of vcgC + V. vulnificus was observed in July 2017. However, there was no significant difference between the years (Fig. 6).

TABLE 4.

Average distribution of Vibrio spp., phytoplankton, and abiotic parameters in Rehoboth Bay water samples

Variable	Avg by yr		P valuea
	2017	2018
V. parahaemolyticus (tlh+)b	542.82	265.96	3.00 × 10−3
V. parahaemolyticus (tdh+)b	8.30	2.43	0.03
V. parahaemolyticus (trh+)b	8.17	3.46	0.18
V. vulnificus (vvhA+)b	188.60	138.68	0.89
V. vulnificus (vcgC+)b	5.13	6.40	0.47
Phytoplanktonc	27,630	48,538	0.05
Tempd	25.60	21.66	0.07
Salinitye	28.22	21.18	<1.0 × 10−3
Dissolved oxygenf	4.30	3.63	0.83

^aBoldface entries indicate significant P values.

^bMPN mL⁻¹.

^ccells L⁻¹.

^d°C.

^eg/kg.

^fmg L⁻¹.

FIG 6.

Distribution of Vibrio spp. in the Delaware Inland Bays from 2017 to 2018. Shows bar plots of V. parahaemolyticus (tlh), V. vulnificus (vvha), pathogenic (tdh+ and trh+) V. parahaemolyticus, and pathogenic (vcgC+) V. vulnificus. The bar plots summarize means for Vibrio genes. Additionally, standard errors bars were calculated using R studio.

Correlation analysis of Vibrio spp. and nutrients.

A correlation analysis of the 2017 to 2018 Torquay Canal and Sloan Cove nutrient and Vibrio data showed that at Torquay Canal, total tlh + V. parahaemolyticus had a moderate positive correlation with TDP, TDN, and PO4⁻³, with a coefficient value greater than 0.5 and a P value of <0.05. Trh +V. parahaemolyticus had a moderate positive correlation with TDP with a coefficient value greater than 0.5 and a P value of <0.05. At Sloan Cove, total V. vulnificus (vvhA+) had a positive correlation with NO₃:NO₂, with a coefficient value greater than 0.5 and a P value of <0.05. All P values were corrected using a Benjamini and Hochberg adjustment to avoid false positives.

Modeling the relationship between Vibrio spp., HABs, and environmental factors.

We used a generalized linear model to determine which explanatory variables best described the detection of V. parahaemolyticus (Table 5). Principal component 1 (PC1) of environmental factors (NO₃:NO₂, TDN, and PO4⁻³) contributed the most in detecting total (tlh+) V. parahaemolyticus (P < 0.05) (Table 5). G. instratium contributed the most in detecting (tdh+) V. parahaemolyticus (P < 0.05) (Table 6). We additionally ran the model using trh+, vvhA+, and vcgC+, but there was no significant relationship discovered (see Fig. S1 to S3 in the supplemental material).

TABLE 5.

Model comparison of generalized linear models with Tlh as the response variable

Model no.	Explanatory variable	AICca	Delta	wtb	RMSEc
M6	Tlh ~ PC1ENVId	101.1	0.00	0.226	0.968
M100	Tlh ~ N023 + PO4 + TDP	101.3	0.24	0.200	0.898
M28	Tlh ~ PC1ENVI + HIe	101.9	0.81	0.234	0.943
M97	Tlh ~ N02:3	102.1	0.97	0.140	0.982
M29	Tlh ~ PC1ENVI + GIf	102.4	1.26	0.187	0.949
M27	Tlh ~ PC1ENVI + HI + GI	102.9	1.84	0.140	0.919
M1	Tlh ~ PC1ENVI + PC2ENVIg + PC1HABh + PC2HABi	104.2	3.15	0.073	0.897
M94	Tlh ~ Sal	107.0	5.95	0.012	1.05
Null	Tlh ~ 1	107.2	6.12	0.016	1.09
M95	Tlh ~ Temp	108.8	7.68	0.006	1.08
M31	Tlh ~ GI	109.1	8.01	0.005	1.09
M30	Tlh ~ HI	109.6	8.52	0.004	1.09

^aAICc, Akaike’s information criterion. A lower AICc indicates a better model.

^bwt, weight.

^cRMSE, root-mean-square deviation.

^dPrincipal component 1 of environmental variables.

^eH. akashiwo.

^fG. instratium.

^gPrincipal component 2 of environmental variables.

^hPrincipal component 1 of harmful algal species.

ⁱPrincipal component 2 of harmful algal species.

TABLE 6.

Model comparison of generalized linear models with Tdh as the response variable

Model no.	Explanatory variables	AICc	Delta	wt	RMSE
M36	Tdh ~ GIa	56.9	0.00	0.339	0.506
M31	Tdh ~ PC1ENVIb + GI	58.8	1.89	0.132	0.500
M32	Tdh ~ PC1ENVI + PC2ENVIc + PC1HABd + PC2HABe	59.0	2.03	0.123	0.461
M30	Tdh ~ N02:3	59.2	2.32	0.106	0.523
Null3	Tdh ~ 1	59.5	2.59	0.093	0.544
M35	Tdh ~ HIf	61.3	4.37	0.038	0.540
M34	Tdh ~ Temp	61.4	4.46	0.036	0.540
M27	Tdh ~ PC1ENVI + HI + GI	61.5	4.59	0.012	0.500
M29	Tdh ~ PC1ENVI + HI	61.5	4.62	0.034	0.500
M26	Tdh ~ PC1ENVI	61.6	4.65	0.033	0.541
M33	Tdh ~ Sal	61.8	4.90	0.029	0.541
M28	Tdh ~ N023 + PO4 + TDP	62.3	5.36	0.023	0.505

^a G. instratium.

^bPrincipal component 1 of environmental variables.

^cPrincipal component 2 of environmental variables.

^dPrincipal component 1 of harmful algal species.

^ePrincipal component 2 of harmful algal species.

^f H. akashiwo.

DISCUSSION

Torquay Canal is a dead-end canal, which is common in residential communities close to bays. They are poorly flushed and subject to anthropogenic inputs from homes and other developed areas. Therefore, it was not surprising that its higher nutrient levels and low dissolved oxygen were significantly different from those measured at the well-flushed Sloan Cove. As dead-end canals are also conducive to harmful algal bloom formation (8), these findings were consistent with a higher total dinoflagellate biomass at Torquay Canal than that at Sloan Cove (Fig. 4).

Dinophysis acuminata has been found frequently at Torquay Canal, and favorable growth is associated with elevated nitrogen constituents, particularly NH₄⁺ levels (15, 16), consistent with our data (Fig. 2). Additionally, this taxon is also an obligate mixotroph that must sequester chloroplasts from cryptophytes by ingesting ciliates (e.g., Mesodinium spp.) that have preyed on cryptophytes (17, 18). Interestingly, we did not observe any M. rubrum or cryptophytes in 2018. Either D. acuminata ingested all available M. rubrum prey prior to our sampling dates in 2018 or D. acuminata can obtain chloroplasts from other phytoplankton.

In June and September of 2017 at Torquay Canal, we also observed high levels of C. subsalsa and H. akashiwo with concentrations ranging from 2.0 × 10⁵ to 6.0 × 10⁵ cells L⁻¹. They are species found commonly in the Delaware Inland Bays (19) and can be both mixotrophic and autotrophic. C. subsalsa blooms are known to co-occur with C. toxicum in the Delaware Inland Bays (20, 21), as was as also seen in our PCA (Fig. 6).

Sloan Cove did not exhibit as high concentrations of HABs as in Torquay Canal, likely due to the lower nutrient levels. G. aureolum, the most commonly reported bloom-forming dinoflagellate in temperate waters (22, 23), was the major species found throughout the 2-year study. HAB species, such as K. veneficum, were detected but at low levels. PCA plots revealed associations between K. veneficum and M. rubrum, which may be reflective of feeding on similar prey items, such as cryptophytes (24–26). In 2017, we detected the presence of both C. subsalsa and H. akashiwo but not in 2018. This finding is likely due to low concentrations of prey items (data not shown) (27, 28). Despite that fewer HAB events occurred in Sloan Cove than those in Torquay Canal, caution should be exercised and routine monitoring continued. Occasional blooms are possible when a “window of opportunity” exists for certain species to outgrow competitors and when there are few zooplankton grazers present (29).

The development of a site-specific Vibrio predictive model based on nutrients and resident phytoplankton may be useful for evaluating water quality in oyster aquaculture areas (2). When sampling sites were compared, similar concentrations of total V. parahaemolyticus and V. vulnificus were recorded, despite differing nutrient and dissolved oxygen concentrations. Whether V. parahaemolyticus and V. vulnificus prefer different environmental conditions remains unclear. The influence of geographic location is unclear. Some studies suggest that this is the case (30, 31), although we did not observe a significant difference between sites, which is similar to other work (32, 33). The levels of both species during the warmer months, though, were consistent with those reported previously (34–36).

V. parahaemolyticus is a moderately halophilic bacterium (34, 37, 38) and was present at higher levels in the higher salinity year of 2017 (Table 1), although the relationship was weak (r = 0.032 and P = 0.11) A significant positive association was observed between total V. parahaemolyticus and TDN, PO₄⁻, and TDP (Fig. S3). Davis et al. also observed a positive association between PO₄⁻ and V. parahaemolyticus in developed areas (34).

Limited work has explored what factors influence the presence of pathogenic strains of V. parahaemolyticus and V. vulnificus. Previous studies have suggested that chlorophyll a levels contribute to higher pathogenic V. parahaemolyticus (tdh+ and trh+) in the water column (31, 39). This suggestion was consistent with our observations, where trh+ V. parahaemolyticus was present in a higher prevalence and concentrations at Torquay Canal, where there is a significantly higher HAB occurrence and biomass than those at Sloan Cove. Additionally, we observed a positive association between trh+ V. parahaemolyticus and TDP, which has not been reported previously. According to the generalized linear model (Table 5), PC1 environmental variables that included temperature, dissolved oxygen, salinity, TDN, PO₄⁻, TDP, NO₃:NO₂, NO₂⁻, and NH₄⁺, contributed the most in the detection of total tlh+ V. parahaemolyticus. G. instratium contributed the most in the detection of total tdh+ V. parahaemolyticus.

Observations from Japan (40) and the Atlantic Coast of the United States (41) have shown that the prevalence of vcgC+ V. vulnificus can vary depending on the system. A correlation between phosphate and vcgC+ V. vulnificus has been reported for the Chesapeake Bay (42). We observed a weak correlation between total (vvha+) V. vulnificus and phytoplankton, which included both dinoflagellates and raphidophytes. This result was unexpected, as prior reports have suggested a strong relationship between total V. vulnificus and both nutrients and HABs (43, 44). Thus, any relationship between V. vulnificus and phytoplankton might be species specific and be influenced by specific nutrients. Future studies will be necessary to derive these potential interactions.

Future studies in Rehoboth Bay should incorporate chlorophyll a data into model development. Vibrio spp. are known to bind to the surface of dinoflagellates and raphidophytes (7) and degrade their surface polymers, such as chitin and plant/algal polysaccharides. It is also possible that pathogenic Vibrio spp. are more abundant in environments where there is enhanced competition with other marine bacteria and phytoplankton for nutrients. In conclusion, this study sought to characterize the relatedness between Vibrio spp. and HABs, as well as determine the influence of nutrients on pathogenic Vibrio spp. and HAB abundance in Rehoboth Bay. We discovered that although the total numbers of Vibrio spp. are similar between Torquay Canal and Sloan Cove, potentially pathogenic Vibrio spp. are more likely to be found at Torquay Canal. Thus, total Vibrio counts at a location may not be sufficient for the accurate prediction of pathogenic potential when considering the placement of aquaculture sites.

MATERIALS AND METHODS

Sample collection and phytoplankton enumeration.

A total of 55 water samples were collected from Rehoboth Bay in 2017 to 2018, including 29 from Sloan Cove and 26 from Torquay Canal for vibrio and phytoplankton analysis. Surface water samples were collected weekly in 1-L amber glass bottles at Torquay Canal (site 1) and Sloan Cove (site 2). These areas were chosen based on their proximity to oyster aquaculture sites, differences in water quality based on the State of the Delaware Inland Bays report, and accessibility by vehicle (1). Station depth ranged from 5 to 7 feet, and surface temperature, salinity, and dissolved oxygen were collected using a YSI 85 (Yellow Spring Instrument Co., Yellow Springs, OH). At site 1, Torquay Canal water samples were collected from July to October in 2017 and April to October in 2018. At site 2, Sloan Cove water samples were collected from May to October in 2017, and April to October in 2018 (Fig. 1) (45). An additional 50 mL of water was collected from all sites and preserved in Lugol’s iodine solution for phytoplankton identification and enumeration via microscopy using a Zeiss IM35 inverted microscope with phase contrast and bright field illumination (46).

The estimated bio-volume of all dinoflagellates was obtained using geometric shape equations from the Baltic Sea environmental proceedings no. 106 and converted to biomass using a conversion factor of 0.760 pgCm⁻³ (cellular carbon) (47). Dinoflagellate biomass as calculated included D. acuminata, G. aureolum, G. instratium, S. trochoidea, and P. minimum. The biomass of K. veneficum was calculated separately because of its small size compared with other dinoflagellates in this study.

Quantification of V. parahaemolyticus and V. vulnificus.

Water samples were also processed for V. parahaemolyticus and V. vulnificus abundance using the three-tube MPN method following procedures described in reference 48. Totals of 100 mL, 10 mL, 1 mL, 100 μL, 10 μL, and 1 μL of undiluted sample were inoculated in triplicate into 10 mL of alkaline peptone water (APW) broth and incubated overnight at 35°C.

Vibrio sp. analysis using real-time PCR methods targeted the species-specific gene thermolabile hemolysin (tlh) to confirm the abundance of V. parahaemolyticus and the species-specific gene Vibrio vulnificus hemolysin A (vvha) to confirm the abundance of V. vulnificus in water. Thermostable direct hemolysin (tdh) and thermostable related hemolysin (trh) were targeted to determine the presence of virulence genes in V. parahaemolyticus. Virulence correlated gene (vcgC) was targeted to determine the presence of the virulence gene in V. vulnificus. Primers, probes, and internal controls were as described previously (49–52).

Real-time PCR was conducted using iTaq universal supermix (Bio-Rad Laboratories, Hercules, CA), as described (51). Real-time PCR cycling was conducted using a Bio-Rad CFX96 real-time system with an initial denaturation/polymerase activation of 95°C for 180 sec, followed by 45 cycles of 95°C for 5 sec, and an annealing temperature of 62°C for 45 sec (53, 54).

Nutrient analysis.

A total of 120-mL water samples from both sites were collected for nutrient analysis measurements for 2017 and 2018. Water samples were filtered through a Millipore 0.2-μm mixed cellulose ester membrane, and filtrates were stored at –80°C. Concentrations of total dissolved nitrogen (TDN), total dissolved phosphorus (TDP), nitrate/nitrite (NO₃:NO₂), nitrite (NO₂-), ammonium (NH₄⁺), and ortho-phosphate (PO₄³⁻) were analyzed by the Horn Point Analytical Lab in Cambridge, Maryland, using a Technicon autoanalyzer II and NAP software (55, 56).

Statistical analysis.

All Vibrio sp. and harmful algal counts were log transformed, and the differences between HABs, Vibrio spp., nutrients, biomass, and environmental parameters were evaluated using a Mann-Whitney rank sum test. A Fisher exact test was used to determine if there was a difference in the number of occurrences for and HABs at each site. The Ggplot2 package in R studio was used to generate a visual representation of the distribution of HABs, Vibrio spp., and environmental parameters. A Spearman correlation test and a Benjamini and Hochberg P value adjustment were used to assess the association between Vibrio spp., nutrients, HABs, and environmental parameters. Correlograms were generated using the Corrgram software package in R. Harmful algal species were transformed using Hellinger, and environmental variables were standardized using Z-scoring for principal component analysis (PCA). A visual representation of PCA was used to determine the relationship between environmental factors and to narrow down specific groups of harmful algal species to use in our model. For HAB PCA, Chloromorum toxicum, Scrippsiella sp., and Gymnodinium instratium contributed over 14% in PCA1 and Karlodinium veneficum and Heterosigma akashiwo contributed over 25% in PCA2. For Environmental PCA, PO₄⁻³, TDP, and TDN contributed over 18% in PCA1, and NO₂⁻ and NO₃:NO₂ contributed over 20% in PCA2.

A generalized linear model was used to describe the relatedness between Vibrio spp., environmental variables, and harmful algal species. To simplify the model, principal component 1 of environmental variables was used in the model instead of each individual variable. H. akashiwo and Gyrodinium instriatum were chosen as representatives for the groups that were formed in the PCA plots because these two species were found regularly. MuMIn package in R was used to calculate Akaike information criterion (AICc) to rank the models. The response variable is tlh + V. parahaemolyticus, and the fixed variables were principal component 1 of environmental variables, principal component 2 of environmental variables, principal component 1 of harmful algal species, principal component 2 of harmful algal species, and H. akashiwo and G. instratium concentrations. Models included both abiotic and biotic variables. All analyses were conducted using R studio 3.3.0 (57).