Okay, so, spoiler alert from the get-go: designing research experiments is really hard, and my initial attitude to the whole research project was messed up. Ah, the hubris of youth. Now, a couple of years after I initially formulated my research plan, I know better - although my experimental design is still not fully cemented. Let's begin from the beginning.
One day*1 I understood that much of what I'd learned about ecologically inspired planting design lay on a shaky ground. This was especially true for may of the main principles and most of the specific methods taught in books on designed plant communities (DPC):
Utilizing detailed plant knowledge to create site-optimized plant lists? Unfortunately, the available detailed plant knowledge turns out to be gappy, contextual and based on undefined sample sizes.
Plant sociability categorization ála Hansen & Stahl *2? Often misunderstood by practitioners, and with very little scientific basis to it.
Role-setting systems, like the ones used by Bund Deutscher Staudengärtner or described as "layers of a plant community" by Rainer & West (2015)? I found almost 10 of them, and they all operate on different rules.
The only way forward, then, would be to start - or rather, continue - research on designed plant communities.
Approaching planting design through the concept of designed plant communities has a lot of promise; in any case, I have yet to find another planting design framework with similarly high ambitions on delivering ecologically grounded, multifunctional and rationally manageable ornamental vegetation for urban environments. That is why I initially thought that it would be possible to tear the theory surrounding designed plant communities apart, sort it into neat categories, pour in some complementary theory from plant ecology, forestry and agriculture, and then reassemble the pieces in each category to create theory on designed plant communities that actually could be tested. For example, an Improved Garden Habitat System could be tested by checking if detailed considerations of plant needs and site attributes would yield healthier, more easily manageable vegetation than a randomized plant selection approach. Or if a Revised Plant Sociability classification would help to design planting patterns that provide denser, more resilient canopy cover and more stable species assemblages.
Unfortunately, especially plant ecology showed to be quite useless in reforming and updating theory on designed plant communities, since plant community ecologists cannot even agree on the definition of a "plant community"*3! Realizing the flaws in both the DPC-approach and my initial research plan led me to a bit of a slump *4. Is there a point to using the DPC-approach at all, when it's so flawed? And then it clicked. IS there a point? That is something to study! The next idea for the research design, then, was to compare plantings based on ideas within the DPC-paradigm with other alternatives. Monocultures of low shrubs were a given point of comparison, their hegemony in Nordic urban green uncontested. Since I was planning to compose my designed plant communities in the research on herbaceous perennials, it was logical to make a counterpart to them that would consist of widely used, "all-round" ornamental perennials. The final point of comparison was initially meant to be lawn, which we only relatively late in the process swapped out with native herbaceous perennials for an "urban meadow" type of approach. These would be compared with each other on the basis of their performance: Their growth and internal vegetation dynamics, their contribution to the technical functions of the planting, their capacity to provide food and habitat for other organisms, as well as how they're perceived by the public. Performance would also be related to the management efforts spent on each type of planting.
If you're familiar with research activities, you will immediately see that each of the points above could be studied for years, from multiple different angles. Additionally, if you're a landscape/ green area practitioner like me, you'll likewise understand that the factors above mostly become interesting when seen in a long-term perspective, since the 3-4 years planned for the experiment initially only cover the early vegetation establishment. As such, it was decided that the experiential questions and biodiversity support would be left for later, in case I would be able to continue my project for another 2-4 years. Technical questions were set aside, as they would require a completely different setup than what we had thought. The first years of the experiments would only focus on vegetation development during the establishment period, where management is at its most intense.
As a design practitioner I was especially interested in researching how design plant communities would fare in tough urban conditions, and thus decided to have the experiments "in real life". In order to get better data, I hoped to be able to have experimental sites on several different locations, either in different cities or different types of places in the same city. What all of the sites should have in common is that they should all include rain gardens, where I could place my experiments. The rain gardens within each site should be as close to identical as possible in their site conditions for improved comparability. This would be important even when approaching the experimental sites as parallel case studies, rather than "experiments" in the strict sense of the word.
I chose the plants for the non-DPC plantings with formalized procedures. These will be the same on each site, that is to say that they will have as many duplicates as there are different sites. At the beginning, I thought I would make the comparison as "fair" as possible, and choose plants with as wide tolerance amplitudes as possible. In practice, this would have required a lot of work (there are more than 1500 taxa - species, subspecies, cultivars and so on - of herbaceous perennials easily available for procurement in Sweden), as there are no simple or certain sources where plants could be filtered by site condition tolerance *5. Instead, I opted to choose both the shrub and the ornamental herbaceous perennials through a popularity contest, that is to say, by collecting sales data per taxa from Swedish plant nurseries to determine which plants have been most widely used in Sweden for the past few years. I collated the data from 5 sources, and found a shrub that outsold its main competitor by an overwhelming margin, as well as 7 herbaceous perennials that repeated in the top 10- top 20 of all of the studied nurseries' most sold-lists.
For meadow plants I could basically have ran the list of the 60-70-odd taxa of native meadow plants available as plugs through a veritable scientific source - maybe by consulting the TRY plant trait database to assess the plants' site condition needs, and base my plant choices on the 7 factors considered in the Ellenberg indicator value system *6. Instead, I copied out the plant lists for different commercial meadow seed mixes from the two large Nordic meadow seed suppliers, and made notes on all of the site-tailored mixes each species appeared in. Then I coded the mixes based on the soil moisture, light conditions and salinity level they were formulated for. This way I could get the suppliers' view on the tolerances of the plants they were selling. In the end I counted how many different light, moisture and salinity conditions each of the plants was recommended for, and translated this to a scale of "wide tolerance-medium tolerance-narrow tolerance" for light and soil moisture conditions, and whether or not the plant is assumed to tolerate salt. I had originally assumed that this would provide me with a clear, usable top 10 list, but due to a variety of reasons the final selection became quite complicated. To save you guys from boredom*7, I'll just say that most Nordic native meadow plants have relatively narrow site condition tolerance amplitudes, which was quite predictable based on their evolutionary histories and distribution patterns. In the end, I wound up with a functional selection of 15 native meadow species, both grasses and forbs.
And then, the designed plant communities. To me, the key to designed plant communities is to think of them as site-specific solutions, both in their ecological and cultural makeup. So I started by making a thorough site analysis of the first "real life" experimental site we got: Annual precipitation, planting bed construction, wind, light, natural and cultural context, length of the growing season… The site analysis alone took me 3 days, because some of the information was quite hard to find, or hard to interpret in a practical manner. After that, I sat down to translate the base information into the Garden Habitat system to more easily make my initial plant lists - similarly to my usual process. On this site I found that the light-, and soil moisture conditions could be interpreted more optimistically or pessimistically: There could be a decent amount of daylight in midsummer (half-sun, roughly corresponding with the garden habitat Woodland Edge), or upcoming buildings and street trees might make the site more shaded (half-shade to shade, typical for Woodland). The biochar-compost mix in the macadam-based substrate might hold water well enough during periods with regular precipitation patterns (mostly normal to occasionally dry soils), or it might dry out very fast and be pretty consistently dry. Due to these, I could basically deduce that the relevant garden habitats are dry or dry to normal woodland (W1, W1-2) and dry or dry to normal Woodland edge (WE1, WE1-2). By mixing and matching the garden habitats and by leaning on some aspects over others, I formulated 3 different planting mixes based on the three most likely interpretations of the relevant Garden Habitats. Besides the Garden Habitats and general climate restrictions, two narrow criteria were applied: the plants should prefer or at least tolerate warm sites, and the planting should mainly consist of plants within the 30-100 cm height range to be voluminous enough to appreciate, while not so high as to reduce experienced safety. The latter I factored in while I was doing my initial picks, and double-checked later. Plant height was one of the very first factors I checked, as I tried to learn as much about each of the taxa as necessary to make informed and well-motivated choices. In the end, I used seven (7!) different sources to fill in the information needed on each taxa. This was partially because I wanted to cross-check the info on each plant - even among the three German sources I used there were dissenting opinions on the proper classification and behaviour of each taxa. Partially I needed the Swedish perspective on things like expected height and blooming time. And then there are the questions of longevity, which thus far have best been compiled by Noël Kingsbury (in his PhD-thesis, as well as in Oudolf & Kingsbury 2015). It may not come as a surprise to you that this also took me several days, almost a week to do.
It was difficult not to "cheat" in the selection process, more so than I anticipated. I realized again how heavily I rely on my tacit knowledge in planting design: "That teacher said that this cultivar is better." "The Grand Old Man forbade us from using this plant." "They -say- that it only tolerates these kinds of conditions, but I've seen this plant in more extreme situations as well…" and so on. Not being able to use tacit knowledge made the process much slower, since I usually apply it heavily already in my first round of picks. On the other hand, both for research as well as for practice, it is important to acknowledge the large influence tacit knowledge has on planting design. This is something I really miss when I read previous research involving planting design, especially in articles written by experts on designed plant communities. I hope that my rigorousness will encourage future researchers to do the same. But as a practitioner, I am unlikely to make my design process this much more detailed - especially since it's quite complex already as is.
In the end, I managed to create 6 planting mixes that I'm quite happy with. If/ when I get more experimental sites, I will have to make new site analyses, which will most likely result in a need to design more DPC. And that is good, since a big part of what I want to communicate with my research is that there are no standard solutions in planting design. Or, well - if the experiments instead suggest that a monoculture of Lonicera caerulea var. kamtschatica 'ANJA'® E ('BLo D'PBR) will always perform best, I guess I'll just have to accept that. Or design new experiments to try to disprove my previous results. As one does.
1* Metaphorically speaking; please forgive me for condensing about 3-4 years of pondering.
2* Deserves it's own post really, but basically it's a way of recommending ideal group sizes for plants when setting them out on site. The recommendations are based on a combination of factors, including observations of species' behaviour in nature, inherent plant traits such as spreading habit, phenology and size, management considerations, and visual effect. Unfortunately, as far as I know there is no publicly available data on how these different criteria are applied to determine the final simplified classification of a plant in any given category.
3* Undeserved slander of plant (community) ecologists aside, it should be obvious that there is a lot that planting designers can learn from plant ecology. The problem is rather that the field is extremely vast with several philosophical camps and wide rifts between different subgroups. So for every interesting point made in a rigorously studied academic paper there are at least 3 different counterpoints (also very interesting, and also presented in very rigorously studied academic papers), and navigating between them would require an in-depth knowledge of the field that would take at least a decade to build. I have 4 years available to me.
4* "I'm in despair!"
5* If you work with Planter or Plantarum or a similar service- I'm sorry, your databases are fine, but I would still need to cross-reference them with a bunch of other sources and do some interpretive dance to figure out how to prioritize between contradictions in the available information.
*6 Before you ask me "Ella, why didn't you use the TRY database from the get-go?": The database (version 5) consists of 11,850,781 trait records and 279,875 plant taxa. Despite this, it lacks information on many ornamental taxa, especially cultivars, which makes it more difficult to use for planting design purposes. Additionally, due to the enormity of the database, it is so.slow. to use, so you really want to know exactly which taxa you want to extract data on before wasting a bunch of time. And, once again - no shade to the TRY database, I think they're doing extremely valuable work, it just wasn't the ideal tool at the start of my project. I did request data from them on the final selected taxa at the end of the process, though.
*7 The full selection protocol will be available for fellow researchers and masochists as a supplement to the future paper.
Comments