Assessing levels of airborne bioallergens in New York City.

Sampling of airborne ragweed and other pollen grains in the City of New York is described as it took place during the months of August and September 1973. Three sampling stations participated in the program. This effort was meant to serve as the forerunner of a citywide volunteer pollen sampling network based on the cooperation of existing health care providing institutions, schools, and stations of the operational sampling network of the New York City Department of Air Resources. Monitoring stations were set up on the roof of Cooper Union School of Engineering at 51 Astor Place in downtown Manhattan, at the Boyce Thompson Institute for Plant Research, Inc. at Yonkers, N.Y. and use was made of the data supplied by Middlesex General Hospital, New Brunswick, N.J. Seasonal results of ragweed pollen sampling, undertaken at the downtown New York City location of Cooper Union with the swing-shield intermittant rotoslide sampler, are similar to those of studies undertaken in the 1940's and 1950's with the Durham slide sampler in the New York metropolitan area. Comparison of the collected data with converted data obtained in 1979 at Long Island Jewish-Hillside Medical Center with the aid of a Durham gravity slide sampler shows the trend similarity of the data collected by us in 1973 and those obtained in 1979. Collected data of the 1973 ragweed season were supplied to the New York City news media on a daily basis, as a much appreciated service to the public.

Bioallergens are part and parcel of the perennially present aero-allergens, especially in a city like New York where a large number of citizens suffer through the various pollen and mold seasons year after year. The ever increasing number of querying phone calls to the Bureau of Science and Technology of the New York City Department of Environmental Protection induced the authors to investigate this problem together with local institutions who volunteered their assistance. Pollen grain and mold pore sampling stations could provide a local inventory of airborne allergens, and an indication of their dispersal throughout the metropolitan area. In conjunction with weather data, this may allow the development of a predictive model for airborne allergen levels. Such a monitoring system in combination with a prognostic capacity would potentially be beneficial to the health of the population of the City of New York.

January 1981
Hay fever may seem rather innocuous. However, when asthma develops as a sequela, which may occur in a large portion of the population suffering from hay fever, mortality may increase appreciably (1,2). Hence, a number of sampling methods were assessed in order to select the most suitable and accurate instrument to serve as a basis for a dependable sampling network: For many years, the Durham or gravity slide sampler has been in use during the ragweed pollination season for monitoring airborne pollen levels, as recommended by the Pollen and Mold Committee of the American Academy of Allergy. Counts were usually taken over a 24-hr period and reported as grains per square centimeter. However, these counts are not necessarily representative due to the influence of atmospheric turbulence and wind speed on slide deposition. Also, the orientation of the slide with respect to the wind direction greatly influences the number of pollen grains trapped. Furthermore, it is impossible to determine the volume of air sampled in the field. Hence, a number of more quantifying sampling methods were investigated. As the most convenient and representative mode of monitoring levels of airborne bioallergens, the swing-shield intermittent rotoslide sampler (3)(4)(5) was selected. This monitor, its action, the handling of the samples it takes, and the conversion of pollen counts, obtained from the rotoslide sampler, to volumetric measurements, assuming the efficiency of the sticky slide edge surface to be 64% for airborne ragweed pollen, have been described fully (4). The efficiency is determined from a measurement of estimate of mean wind speed during the sampling period, and varies linearly from 68% at zero speed to 49% at 10 mps for ragweed pollen. Volume monitored is determined from length of sampling period, and a constant which includes the cross sectional area of the slide-edge sampling surface, the linear travel per revolution and rate of rotation. The normalized pollen count is divided by the volume and efficiency to determine concentration of pollen grains per cubic meter of air.
In New York City and surrounding regions (6), three pollen seasons can be distinguished, that of early spring, early summer, and of late summer. The earliest pollinators are elms, maples, and willows in April. Clinically more important are the poplars and birches which shed their pollens in late April and the first two or three weeks in May. Oaks, ashes, and beeches also pollinate in May, hickory a little later. Grass pollen concentrations, though not as high as those produced by trees, affect more people. with often more severe symptoms, in the early summer approximately until early July. Nonetheless, we have found grass pollens through September. Pollen of the weed English plantain can be found from May through August. The prime offender of pollinosis, however, is ragweed pollen, which pollutes the New York City air in bothersome quantities from approximately mid-August to the end of September.
The various pollens and molds not only show seasonal patterns but also pronounced diurnal emission cycles can be observed. Each genus has its own characteristic diurnal emission pattern (7). Ragweed pollen emission usually begins near sunrise, peaks a few hours later, and decreases through the afternoon. The peak count most frequently is between 9 and 10 AM. Periods of maximum airborne pollen concentrations are directly related to maximum atmospheric turbulence. The process of pollen release takes place in two stages: (1) release from the anthers is determined by time of day, temperature and relative humidity; and (2) the actual transfer of the pollen grains from leaf surfaces to the air takes place by turbulence and windspeed. These processes are as follows: Stable atmospheric conditions usually prevail in the early morning hours before sunrise and are associated with a low-level temperature inversion, which means 172 very low surface wind speeds with complete absence of turbulence with isothermal to adiabatic lapse rate conditions. As the sun rises higher and heats the surface, unstable conditions develop, accompanied by superadiabatic lapse rates which promote turbulence. This facilitates pollen transport from the plant leaves to the air. Emission most typically begins when stable, moist night time conditions give way to unstable and drier air shortly after sunrise (7). Vertical pollen profiles show great variability from day to day because of varying weather conditions.
In this respect, the efficiency of the intermittent rotoslide sampler, used in our three-station network, is independent of wind direction and windspeed. Raynor, Ogden, and Hayes (8) sampled ragweed pollen concentrations at heights of 1.5, 11.3, 22.8, 45.7, and 108.2 m, respectively, with the intermittent rotoslide sampler. They found that the upper level was the most representative of regional concentrations and the least influenced by local turbulent conditions. Actual ragweed pollen grain concentrations did not differ much with the presence or absence of nearby local sources. The authors concluded that "the diversity in size, distance, and location of ragweed sources, and in timing, duration, and magnitude of pollen emission, plus the variability in meteorological conditions between source and sampling locations, form such a complex set of possible variables that variation with height over short periods cannot be related to parameters measurable at the sampling site. However, when averaged over longer periods of time, these variables tend to counteract each other so that a smoothed profile results."

Sampling Methods
In view of the above, we decided to launch a city-wide network of pollen and mold sampling stations equipped with the intermittent swingshield rotoslide sampler developed by Raynor and Ogden (3)(4)(5). Two instruments were purchased commercially while an antiquated model with circular shield in the laboratory of the Bureau of Science and Technology was transformed by laboratory technicians into an intermittent swing-shield rotoslide sampler (Fig. 1). This instrument was installed on the roof of Cooper Union School of Engineering at 51 Astor Place in downtown Manhattan. Sampling for the ragweed season began on August 6,1973 and was terminated on September 26, 1973.
The instrument was timed to sample intermittently for 1 min every 12 min from 6 to 10 AM every day, except on weekends, as usually pollen counts rise Environmental Health Perspectives sharply during early to mid-morning, a rapidly thereafter. In the months previous to the begii ragweed season, the Boyce Thompson Plant Research, Inc. at Yonkers, N. proached and found willing to have rotoslide samplers installed on its pr analyze exposed slides on a daily basi Middlesex General Hospital in New N.J., had been supplying ragweed pc the New York City news media for the years. As these data were derived fror sampler identical to the ones installe Union and Boyce Thompson, we cc hospital to obtain its daily count. Our acceeded to graciously. This permissioi a miniature ragweed pollen sampling three stations.
In the meantime, we had receiv rotoslide samplers and six sampling he New York State Government, in Al While these were readied to be instal wide network of sampling stations, a three-station network consisting of the tions was ready to operate at the begi 1973 ragweed pollen season. The C( location represents the central area oi ing end of the pollen grains. MiddleE data in general gave an indication of rai levels to which the Borough of Manhati exposed. New Brunswick, N.J., is siti west of New York City where the mc pollen pollution occurs during a weath4 January 1981 which a southwesterly to westerly wind prevails.  Table 1 shows the daily ragweed pollen counts at the Cooper Union, Boyce Thompnd decrease son and Middlesex General Hospital stations, together with morning surface wind direction and nning of the morning average wind speed in the mixing layer as Institute for measured at Fort Totten, N.Y. Correspon(ling Y., was ap-weather observations at the U.S. Weather Buone of our reau's Central Park Observatory in Manhattan also operty, and were analyzed. It was found that weather dlata at s. the park could not always explain the collecte(d Brunswick, sampling data, most probably due to the fact that llen data to Central Park is located within a large heat islan(d past several with a varied topography due to an agglomeration n a rotoslide of high rise buildings. and due to the prevailing sea d at Cooper and land breeze system superimposed on the local ntacted the meteorological conditions. request was Table 1 clearly shows a number of peaks in the n resulted in daily ragweed pollen count which will be dliscusse(l. network of On August 10, the wind was southwesterlvy i.e., from areas which are source regions of pollens and( ed a dozen mold spores for New York City, which at the time ads from the was located in the warm zone of a frontal svstem. lbany, N.Y. The same situation prevailed the following day. As [led in a city indicated, the cold front passed the area between preliminary August 12 and 13, when the wind veere(1 to NW. above loca-However, the area northwest of New York Citv inning of the also serves as a source region for pollen grainss. Doper Union Hence, the mildly higher sampling result of 18 n the receivpollen grains/mi of air sampled in the citv and( 1 /ni) sex Hospital at Boyce Thompson. On August 16, raini in the ealyI! gweed pollen morning hours cleansed the atmosphere. Henice. tan would be the pollen count was zero. The probable reason for uated souththe Boyce Thompson count of 4/m' is that the )st offending institute is surrounded by suburban an(d rur al alreas er pattern in which serve as nearby source regions, while the rain stayed south of the station. On August 22, a similar situation developed, only now precipitation also occurred at the Boyce Thompson site. The last days of August show a large increase in pollen sampling counts, peaking August 28 at the three ragweed pollen sampling stations. During this period, the Bermuda High extended over the eastern half of the continental U.S., thereby creating a steady southwesterly atmospheric flow which transported the large amount of pollens from the source regions to the New York City metropolitan area. Subsequently, the high pressure area retreated. Early September a similar situation developed.
On September 6 and 7, the pollen count peaked at Boyce Thompson and Cooper Union respectively. From this occurrence on, counts diminished steadily towards fall with two minor peaks occurring on September 14 and 20. In general, pollen counts increase in the warm sector of a weather disturbance, when southwesterly winds sweep pollens dDuring Labor day weekend, wind direction was SW on 9/1 then was NE on 9/2 and variable on 9/3. eLow pressure area to south of New York City moved eastward from 9/14 to 9/15 bringing rain to area on 9/16 NYC again in warm sector of following frontal system, with SW wind. Cold frontal passage brings NE wind on 9/17. fWind veered to SE with approaching frontal system on 9/22, then to SW on 9/23 after which a cold front passed.
Environmental Health Perspectives 174 and mold spores towards the metropolitani area. Winds from southwest to west-northwesterlv direction usually bring great quantities of allergenis to New York City. Frontal or local convection rainstorms completely clear the atmosphere bv their scavenging action. Simultaneously with analyzing for ragweed pollen presence at Cooper Union, slide edges were scanned for the presence of grass and English plantain pollens, an(d for alternaria, helminthosporium and hormodendrum mol(d spores. Table 2 gives the corresponding (laily counts. High counts of pollen grains and mol(d spores coincide with those of ragweed pollens as is shown graphically in Figure 2. This fact indicates that the pollen and mold source regions for the sampling stations, whose counts are reported here, are located in areas to the southwest to westnorthwest upwind. In the course of this studly, anl endeavor was made to correlate sampling results with weather parameters. A steady relationship could not be established. It was found that wind direction and average speed in the atmospheric mixing layer gave the best indication of the level of pollen and mold spore counts. Analysis of height of the mixing layer showed that this parameter does not seem to have a significant influence on the counts registered.
In order to show that the ragweed pollen data

176
Environmental Health Perspectives FIGURE 2. in 1973. collected in 1973 with an intermittent rotoslide sampler are still representative for the 1979 season, these have been plotted in Figure 3, which also shows the results of seasonal sampling with the aid of a Durham gravity slide sampler, obtained in 1979 at Long Island Jewish-Hillside Medical Center. These counts are lower than those collected with the rotoslide sampler. In this respect, Durham (9) estimated, by means of parallel volumetric measurements, that with his sampling device the amount of ragweed pollen grains, deposited on 3.6 cm' of the microscope slide surfaces (2 x 1.8 cm2), is approximately equivalent to the numbers of grains per cubic yard of air. Hence, we converted the 1979 data, expressed in grains per square centimeter of slide surface, to numbers per cubic yard by multiplying them by 3.6. In turn, we adjusted the results to the number of ragweed pollen grains per cubic meter of air in order to make the data comparable with our 1973 data.

Discussion
At this point, it is useful to elaborate on the process of pollen dispersal from the ragweed flower as described by Holmes and Bassett (10) thereby quoting Bianchi et al. (11). The latter showed that the stage of dehiscence of the ragweed flower is a function of temperature, that is the preparation for the process of dehiscence. After that, the process is a function of vapor pressure in the microclimate of the plant. Flowers will not disgorge pollen until the vapor pressure reaches the threshold which permits the anther envelop to desiccate to the point where it ruptures and allows the pollen to be spilled out. Dehiscence, therefore, is a function of temperature and vapor pressure. There are three steps before actual pollen dispersion takes place: ejection from the flower, adherence of pollens to adjacent vegetation, and flotation caused by turbulent air. The latter only takes place under conditions of a superadiabatic lapse rate in the microclimate of the plant cover, when the relative humidity sharply decreases. The period of maximum pollen concentration varies widely but occurs in the morning hours, except during precipitation. These observations mean that in order to predict possible pollen concentrations at areas of deposition, two sets of atmospheric data have to be procured: micrometeorological lapse rate, temperature, and humidity in the source region and synoptic meterological data in the area of deposition. The air cleaning action of precipitation in both areas is self-evident, so that the forecast of precipitation is of the utmost importance. In order to establish local mean seasonal curves of pollen concentration, Raynor and January 1981 Hayes (12) recommend that several years of quantitative ragweed pollen concentration measurements be obtained for sites of interest. Consideration should be given to type of air mass, length of time this air mass has lingered over pollen-producing regions, frontal passages, air trajectory, wind direction and speed, as well as amount of sunshine, turbulence, especially superadiabatic lapse rates early in the morning in the lower boundary layer at the source. Periods of maximum airborne pollen concentration seem to be related to periods of maximum air turbulence in the boundary layer which together with increasing wind elevates the pollen count sharply. Hence, it should be realized that forecasting the pollen levels at destination is a very complicated matter in which a variety of meteorological and micrometeorological parameters should be available before attempting an indicative correlation. Also, the amount of precipitation in the source region should be considered in the months that the ragweed plant develops, as well as the amount of sunshine in the days that the plants reach maturity and release their pollen.
When comparing the results of the ragweed pollen grain data collected with the rotoslide sampler in 1973 and with the Durham sampler in 1979, it should be realized that the instruments operate on completely different principles. Using the factor of 3.6 to equate sampling results of both instruments is wholly empirical. The only advantage is that trends in the data may be compared. Figure 3 shows that although on some days data are missing, a general trend can be distinguished. Trends registered in 1973 and in 1979 are slightly out of phase, but still cover the same season. The being out of phase of the data may well be due to the fact that in 1973 and in 1979 sampling locations are different geographically by a distance of approximately 10 miles, although still within the confines of New York City. Similar geographical differences in peak occurrence of ragweed pollen grains are mentioned by Wiseman et al. (13). Hence, the authors feel that the 1973 ragweed pollen data are still representative of presently existing conditions.

Conclusion
Sampling for the presence of airborne pollens and mold spores was undertaken in the New York City Metropolitan area. Simultaneously, U.S. Weather Service data for Fort Totten, N.Y. and Central Park Observatory were analyzed. This study would be needed for a number of years, from March through October, to establish yearly curves for the presence of the various types of allergens in the metropolitan area. Micro-and macroclimatological data for the source regions and area of deposition respectively and epidemiological data in the receiving area would complement the gathered bioallergenic data to establish patterns which could be used as a tool in the forecast of airborne allergens over New York City. This information would not only enable allergic patients better to protect themselves against the expected cityward flow of allergens, but it also would be of value to consulted physicians in correlating symptoms with the presence of certain pollen grains and mold spores. The reported investigation was undertaken as a beginning of the planned program. Sampling results of the 1973 ragweed pollen season were supplied to the news media on a daily basis (weekdays only) as an important service to the public.
Ragweed pollen sampling data collected in 1973 and in 1979, although obtained with different samplers and at different geographic locations within New York City, still show a similar trend qualitatively within the same time span. Hence, we submit that the results of our sampling program in 1973 represent presently existing airborne pollen grain conditions in New York City.